Excavation · Where Did SARS-CoV-2 Come From? (2002 to 2026)
Area: Medicine and Health · Science and Nature · Law and Justice · Media and Misinformation · Question: Where and when did it begin?, Is a widely shared claim true? · Starts from a widely shared claim
Source notices. Some sources this page discusses are retracted, corrected or unpublished.
The short answer
Unsettled
Current assessment · as of 2026-10-06
Not settled. Neither a natural spillover nor a laboratory-related incident has been shown: the market evidence is disputed, the lab case is circumstantial, no progenitor is public.
Central claim: The origin of SARS-CoV-2 has not been established: no document or sample read for this dig shows an infected animal, an intermediate host, a
Evidence ●●●○○ Public belief ●●●●○ Source checked: secondary
What would settle it. A progenitor virus (above 99 on that identity scale) found in an animal population, or in a laboratory's records and freezers, would move the weight most. Next: the more than 500 early-pandemic sequences SAGO says were not shared; the early-case line list with onset dates and coordinates; sampling data for the animals sold at the market and its supply chain; and WIV's laboratory notebooks, sequence archives and staff health records, which WHO and NIH requested and did not receive. Not yet read, and listed as sub-questions (claims co-sq-...): both Science errata, the CIA, FBI and DOE primary statements, the BND assessment, the DARPA decision record, the WIV database record, the original January-February 2020 Proximal Origin messages and the June 2026 ODNI release.
Two hypotheses remain open, and each lacks its decisive piece. Natural spillover from animals (directly from bats or through an intermediate host) is supported by the clustering of the earliest recorded cases around the Huanan Seafood Wholesale Market in Wuhan, by SARS-CoV-2 found in 74 of 923 environmental samples taken at the market after it closed, and by the sale of live mammals that can be infected. Against it: no infected animal has been found (457 animal samples, all negative, were taken after closure), statisticians have published objections to the case-geography analysis, the genetic argument for two separate spillovers lost most of its stated strength in a 2023 erratum according to two critics (the erratum itself was not read), and the 2021 WHO-China report itself says no firm conclusion about the market's role can be drawn. A laboratory-related incident is argued from circumstance: the outbreak began in the city where the Wuhan Institute of Virology works on coronaviruses; the US Intelligence Community says WIV researchers performed SARS-like coronavirus experiments in BSL-2 laboratories as of January 2019 and probably did not always use adequate biosafety precautions; the virus has a furin cleavage site that no other known sarbecovirus has; and a 2018 grant proposal (DEFUSE, never funded by DARPA according to the sources read) described introducing furin sites into bat coronaviruses. Against it: the same US report says it has no indication that WIV's holdings included SARS-CoV-2 or a close progenitor and no direct evidence of a specific incident; the proposal is a plan, not a record of work done; and the furin site is not unique to SARS-CoV-2 among coronaviruses. Neither side has a progenitor virus: the closest known relatives score 96.8 (BANAL-52) and 96.2 (RaTG13) on the identity scale on which the US report puts a close progenitor above 99. What institutions say is shown separately because it is adoption, not evidence. Five US intelligence elements (with the National Intelligence Council) lean natural, the FBI and the Department of Energy lean lab and the CIA and one other could not decide (ODNI, 2023; confidence levels are redacted in the public text); the CIA said in January 2025, at low confidence, that a research-related origin is more likely (as quoted by CBS and AP; the CIA document was not read); WHO's advisory group SAGO (2025) says the weight of available evidence suggests zoonotic spillover but that it could not assess a laboratory route because data were not provided; the House Select Subcommittee majority (2024) calls a laboratory or research-related accident likely, and its Democrats say the arguments are largely circumstantial. None of these is an anchor. An HHS debarment of EcoHealth Alliance and its former president is about grant compliance, not about where the virus came from. Several items could not be read and are filed as searched gaps: the Science errata for two key papers, the CIA, FBI and DOE primary documents, the German intelligence service's assessment, the Wuhan database claim, the original early-2020 Proximal Origin messages and the documents ODNI released in June 2026.
Based on: co-origin-not-established co-no-progenitor-public co-zoonosis-hypothesis co-market-pattern co-market-samples co-lab-hypothesis co-odni-2023-wiv-findings co-fcs-unique-among-sarbecoviruses co-defuse-shows-engineered co-assessments-say-open
We give no probability figure. Confidence words describe how sure we are of the basis, and no source supplies a number that would not be invented.
Challenge this assessment A challenge needs a specific claim and evidence we can check. How challenges work.
searched gap confidence provisional · result of a search
The origin of SARS-CoV-2 has not been established: no document or sample read for this dig shows an infected animal, an intermediate host, a progenitor virus, or a laboratory that held, made or released the virus.
Anchor. Read in full or in the passages used (2026-10-05): ODNI 2021 and 2023 ("no confirmed animal source of SARS-CoV-2 has been identified" and "no indication that the WIV's pre-pandemic research holdings included SARS-CoV-2 or a close progenitor"), the WHO-China report 2021, SAGO 2025, both House reports, both Senate staff products, the DEFUSE proposal and drafts, and the Baric transcript. None presents an infected animal, intermediate host or progenitor, or a laboratory record of one. This is a search of the documents we could open; the classified annex to the US report, China's laboratory records and its early data were not available to us.
Would change if. a virus more than 99 percent similar to SARS-CoV-2 is found in an animal population or in a laboratory's records or freezers, or an infected market animal or its supplier is identified
Next step. Obtain the items listed in the discriminating tests: early sequences, market-animal and supply-chain sampling data, and laboratory records (dt-progenitor-search, dt-early-sequences, dt-market-animals, dt-wiv-records).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
A dig on the origin of COVID-19. It is filed as a set of claims about what is known, with the market-origin and laboratory-related hypotheses each tested in plain words at the same standard. The dig keeps apart (1) what studies and samples show, filed as synthetic or primary text; (2) what the primary documents say (the intelligence reports, the proposal, the grant records, the debarment notices); (3) the popular claims built on these, stated in plain words with Yes and No positions; and (4) the institutional record (agencies, WHO, a House report, a Senate staff report, China's white paper), filed as adoption, never as evidence. No claim says either hypothesis is refuted or established.
The gap here runs both ways, and it is visible in the sources. Institutions have said more than the evidence they publish. Parts of the US government and the House majority call a laboratory-related accident likely, while the US report they rely on says it has no indication that WIV held SARS-CoV-2 or a close progenitor and no direct evidence of a specific incident; a White House page states that a lab-related incident "is the most likely origin" without citing evidence on the page. In the other direction, WHO's advisory group says the weight of available evidence suggests zoonotic spillover, while its own report says the market's role is not conclusive, that it was given no raw data, and that it cannot rule a laboratory route out; a 2021 WHO-China report called a laboratory incident "extremely unlikely" on the strength of statements from laboratory representatives, and WHO's Director-General said that assessment "was not extensive enough". Meanwhile the natural-origin side's three main scientific papers each have a published critique or a correction: the case-geography analysis (statisticians' objections), the two-lineage analysis (a 2023 erratum that two critics say cut its Bayes factor to about 4) and the market-sample interpretation (a 2023 reanalysis of the sequencing data). The direct evidence that would settle it is, on both accounts, held by China: early sequences and case data, market animal and supply-chain data, and laboratory records. Stratah records all of this and gives none of the institutional positions evidential weight.
Every point on the timeline links to its source. Open the timeline.
Status: passed with open items · reviewed 2026-10-06. Mechanical pass: python3 build/tools/review_dig.py covid-origins --write. Every check passes except the taxonomy entry, which lives in build/taxonomy.yaml, an owner-only file the author correctly did not edit (log L-08 proposes the line). Conformance on the merge result (branch merged with origin/main, no commit): 0 errors overall; for this subject two warnings only (a key point is 205 characters, over the 170 guide; no taxonomy entry). Sources opened and read by me (not by the author's report): Andersen et al. 2020 (Europe PMC full text: 'not a laboratory construct', 'no laboratory-based scenario is plausible', 'impossible to prove or disprove', the HKU1 sentence); Coutard 2020 (OC43/MERS/HKU1 cleaved at S1/S2); Wu and Zhao 2021 ('occurred independently for multiple times'); Worobey et al. 2022 (the full 'emergence ... via the live wildlife trade' sentence, 155 of 164 cases, 1% contour, medians 4.28 / 5.74 / 4.00 / 16.11 km, p=0.029); Pekar et al. 2022 (BF 61.6 and 60.0, 787 genomes, 18 Nov 2019 (23 Oct-8 Dec), the 10 December seafood vendor with no published genome); Huang et al. 2020 (1 Dec onset, no epidemiological link); Wu et al. 2020 (hospitalised 12 Dec 2019); Liu et al. 2023 (publisher HTML: 923 and 457 samples, 188 animals / 18 species, 74 = 70 + 4, 87.5% (56/64) West Zone, 134/678 shops, three isolates with Ct<30, 'cannot be determined from the analyses available', 'cannot yet be ruled out', note added in proof); Crits-Christoph 2024; Temmam 2022 (96.8% BANAL-52, 96.1% RaTG13, no furin site); Chan and Zhan 2022 (both the 'no known sarbecovirus' and the 'consistent with natural evolution' sentences); GenBank MN996532.1 (submitted 27-JAN-2020, collection 24-Jul-2013, replaced 13 Oct 2020); ODNI 2021 unclassified summary and Updated Assessment (stored copies: four elements and the NIC at low confidence, one element at moderate confidence, three unable to coalesce, 'no confirmed animal source', the genetic-engineering line); ODNI June 2023 (stored copy and the govinfo PDF, read with layout: the three agency bullets, 'no indication ... nor any direct evidence', 96.2 / 96.8 / above 99 percent, BSL-2 as of January 2019, 'probably did not use adequate biosafety precautions', 'consistent with but not diagnostic', the non-respiratory hospitalisation, 'does not address the merits'); WHO-China joint report (6 Apr 2021 PDF: 174 = 100 + 74, earliest onset 8 Dec with no Huanan exposure, 55.4 / 28.0 / 22.6 / 4.8 / 44.6 percent of 168, 'a similar number of cases were associated with other markets', 'No firm conclusion therefore ...', the Likert scale and 'extremely unlikely', the laboratory-representative statements, the 1 Dec / 26 Dec patient, the case definition until mid-January 2020); WHO Director-General's remarks of 30 Mar 2021 (all four quoted passages); SAGO 2025 (the weight-of-evidence sentence, 'cannot conclude with certainty', 'not in a position to rule this out', 'did not have access to original raw data', 'no evidence has been presented, other than speculation', 'not conclusive that the HSM was where the virus first spilled over', 'this has been refused', the two sentences on the US reports and the House majority); SAGO news item of 27 Jun 2025; House SSCP majority report (govinfo PDF: the finding heading, 'weight of the evidence increasingly supports', 'could have created and released', 'initially hid the most dangerous aspect', and that the 'single introduction' point is sourced to a quoted commentator); House SSCP Democrats' report ('largely circumstantial but cannot be dismissed out of hand', 'impossible to draw any conclusions without reviewing additional lab records from WIV', the two section headings); Senate HELP minority staff report ('not intended to be dispositive'); ASM statement of 28 Oct 2022; DEFUSE main volume (Internet Archive OCR: USD 14,209,245, the proteolytic/furin cleavage-site sentences, the human-specific cleavage-site sentence, cover letter 24 March and form field 3/27/18); Baric transcribed interview (Senate PDF: "We've never dropped a furin cleavage site into a coronavirus like that"); HHS stored records (suspension effective 15 May 2024, EHA ineligible through 14 May 2029, Daszak notice of 21 May 2024 and ineligible through 20 May 2029, the WIV 17 Jul 2023 suspension and 19 Sep 2023 ten-year debarment in the ARM); House Oversight debarment release; State Department fact sheet of 15 Jan 2021 (stored copy: the 'reason to believe' sentence); White House page (stored copy and the live page: 'A lab-related incident involving gain-of-function research is the most likely origin of COVID-19', the five numbered items, metadata datePublished 2025-04-18 and dateModified 2025-08-15); House Oversight press release of 2 Dec 2024 ('The FIVE strongest arguments', 'a biological characteristic that is not found in nature'), which is the actual holder of the refuted wording and is attributed as such; China's white paper (all six pages: 'the study on the origins of SARS-CoV-2 conducted in China has ended', the cold-chain sentence, 'Evidence Pointing to the US as the Origin of Covid-19', 457 / 74 of 923, the Los Alamos/NIH/ODNI biosecurity sentence); NHC news item; McCowan 2025 ('the Bayes factor dropped to 4'); Weissman EJW 2026 ('a less compelling ~4.3'); Stoyan and Chiu 2023; Andersen written testimony 2023 ('If convincing new evidence were to be discovered ...'). That is 33 sources opened; every anchor source of every refuted claim, of every evidential_weight 5 claim and of the headline claim was opened, except as listed under what I could not check. Every quotation, figure, date and document number I checked matched the dig's text, including the three quotations the mechanical quote check could not match because no stored copy exists. Judgment checks (REVIEW.md A to F). A: anchors say what the claims say; the one place the dig states a stronger source sentence than a summary would (Worobey's 'occurred via the live wildlife trade') is quoted in full and answered in the same claim. B: every absence claim (co-origin-not-established, co-no-progenitor-public, co-market-animals-untested, co-sq-market-supply-chain, co-sq-wiv-records, co-sq-early-sequences) carries absence_anchor: true and sits at provisional; I found no unmarked absence claim above provisional. anchor_checked: primary appears only where the primary document was read; the headline claim is honestly marked secondary and the headline is therefore draft. Each refuted claim uses narrative_drift, the weakest class, and refutes only the words 'proven', 'shows' or 'not found in nature'; no claim uses fabrication or institutional_propaganda. C: no claim draws weight from how many people or institutions hold a view; the agency, committee, WHO and white-paper claims are filed as adoption records of what a document says and the threads file carries confers_weight: false with explicit firewalls. D: both sides are held to the same bar: the 'proven' claim is refuted on each side (co-market-origin-proven, co-lab-origin-proven), both hypotheses are contested at low confidence, and the strongest item on each side is stated with its own counter-evidence. E: the headline, search summary and assessment state no more than the claims support, carry no hype, and say plainly that the question is open; quotations are short and attributed. F: the log is append-only, records the discrepancies found (L-04, L-05), what could not be read (L-06), the filing decisions (L-07) and the author-side independent check and its nine fixes (L-10). Not checked by me, and why: the two Science errata (Worobey 2024, Pekar 2023) are paywalled, so the Bayes-factor change rests on two critics, as the dig says; the CIA, FBI, DOE and BND primary documents (not public); the Los Alamos and NIH reports China's white paper names; the published JRSSA versions of the two Worobey critiques; the 2023 Bloom correction (an image); the WHO-China annexes; the HHS notice to WIV; the HSGAC Slack volumes, the FOIA DEFUSE drafts, the Johnson letter, the Marshall summary, the Biden and joint statements, the NIH 2020 letter, the OIG audit, the ODNI 2026 page and the CNN, NPR, CBS, Tagesspiegel, NOS and Scientific American pages (each already filed as secondary, unchecked or a searched gap at provisional confidence). Every one of these is marked not read in the dig itself. Verdict posted on PR #13: APPROVE, escalated to the owner because review.yaml is an owner-only path; I did not merge, did not touch build/site.yaml, and changed no other file. The review record.
refuted: the claim below is false confidence high · our conclusion
The furin cleavage site of SARS-CoV-2 is a feature that does not occur in nature.
Anchor. Furin cleavage sites occur naturally in other coronaviruses: Andersen 2020 ("other human betacoronaviruses, including HKU1 (lineage A), have those sites"); Coutard 2020 (present in other coronavirus groups); Wu and Zhao 2021 ("occurred independently for multiple times in the evolution of the coronavirus family"). What is true is narrower: no known sarbecovirus has one at that location (co-fcs-unique-among-sarbecoviruses) [read].
Would change if. the claim is restated as not found in any other sarbecovirus, which co-fcs-unique-among-sarbecoviruses records as established
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
refuted: the claim below is false confidence high · our conclusion
Studies have proven that SARS-CoV-2 spilled over from animals at the Huanan market.
Anchor. Worobey et al.: "While there is insufficient evidence to define upstream events, and exact circumstances remain obscure, our analyses indicate that the emergence of SARS-CoV-2 occurred via the live wildlife trade in China, and show that the Huanan market was the epicenter of the COVID-19 pandemic." Liu et al.: origin "cannot be determined from the analyses available". WHO-China: "No firm conclusion therefore about the role of the Huanan market in the origin of the outbreak, or how the infection was introduced into the market, can currently be drawn." SAGO 2025: "it is not conclusive that the HSM was where the virus first spilled over" [read].
Would change if. an infected animal or progenitor is found in the market's supply chain
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
refuted: the claim below is false confidence high · our conclusion
The DEFUSE proposal shows that SARS-CoV-2 was engineered in a laboratory.
Anchor. The proposal is a plan dated March 2018 that contains no SARS-CoV-2 sequence (the virus had not been described). DARPA did not fund it according to the House majority, the House Democrats and SAGO (no DARPA document was read, co-sq-darpa-decision). ODNI 2023: "no information ... indicating that any WIV genetic engineering work has involved SARS-CoV-2, a close progenitor, or a backbone virus that is closely-related enough". House Democrats: "it is unfortunately impossible to draw any conclusions without reviewing additional lab records from WIV". Baric (2026): asked whether any such work had been done before the application, "No. We've never dropped a furin cleavage site into a coronavirus like that" [read].
Would change if. a record shows work matching the plan was carried out and produced a virus matching SARS-CoV-2
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
refuted: the claim below is false confidence high · our conclusion
A laboratory origin of SARS-CoV-2 has been proven.
Anchor. ODNI 2023: "no indication ... nor any direct evidence"; ODNI 2021 and 2023 say both hypotheses remain plausible; SAGO: "no evidence has been presented, other than speculation from scientific or intelligence reports, that supports a laboratory-related incident"; House Democrats: "largely circumstantial"; Senate HELP minority staff say their conclusion "is not intended to be dispositive" [read].
Would change if. laboratory records or samples show the virus or a progenitor was held there
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
The US Intelligence Community (2021 and 2023), WHO's Director-General (2021) and WHO's advisory group SAGO (2025) each say in their own documents that the origin of SARS-CoV-2 has not been established. The 2023 US report says both a natural and a laboratory-associated origin remain plausible.
Anchor. ODNI 2021: "the IC remains divided on the most likely origin of COVID-19. All agencies assess that two hypotheses are plausible: natural exposure to an infected animal and a laboratory-associated incident." ODNI 2023: "All agencies continue to assess that both a natural and laboratory-associated origin remain plausible hypotheses to explain the first human infection." WHO Director-General, 30 Mar 2021: "all hypotheses remain on the table ... We have not yet found the source of the virus". SAGO 2025: "SAGO cannot conclude with certainty where and when this occurred" and "is not in a position to rule this out as a possibility" (the laboratory route) [read].
Would change if. not applicable (a record of what documents say); it would change only if one of the institutions publishes a finding
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence moderate · what a source says
The joint WHO-China report (2021) counted 174 cases with onset in December 2019 (100 laboratory-confirmed, 74 clinically diagnosed). Of 168 with an exposure history, 55.4 percent had any market exposure (28.0 Huanan only, 22.6 other markets only, 4.8 both) and 44.6 percent had none. The report says "No firm conclusion therefore about the role of the Huanan market in the origin of the outbreak, or how the infection was introduced into the market, can currently be drawn."
Anchor. WHO-China joint report, 6 Apr 2021 file, read: 174 cases reported to the national surveillance system with onset in December 2019; earliest onset 8 December 2019, "no history of exposure to the Huanan market"; 164 resident in Wuhan; other exposures included cold-chain products 26.4 percent and live animals 11.8 percent; seven clusters (15 cases); three suspected cases with onset 1, 2 and 7 December were excluded by the Chinese team. "A similar number of cases were associated with other markets" [read].
Would change if. the Annex E4 case data, the line list and the discarded case records are reviewed jointly, as the report itself recommends, and the counts change
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
Published sources give different dates for the earliest known COVID-19 case in Wuhan, and they measure different things: onset on 1 December (the first patient in a Lancet series), onset on 8 December (earliest onset reported in the WHO-China report, no Huanan exposure), hospitalisation on 12 December (a Nature paper), and a modelled lineage-B infection on about 18 November (Pekar et al.). No single earliest-case date is established.
Anchor. Huang et al. 2020: "The symptom onset date of the first patient identified was Dec 1, 2019" and "No epidemiological link was found between the first patient and later cases". WHO-China report: "the earliest onset date reported to the NNDRS became ill on 8 December 2019". Wu et al. 2020: "since the first patient was hospitalized on 12 December 2019". Pekar et al. 2022: lineage B primary infection "around 18 November 2019 (23 October-8 December)" and an unambiguous earliest case with onset 10 December (a seafood vendor, no published genome) [read]. The WHO report also records a 1 December onset in a man whose further illness began on 26 December. Whether the 1 December patient is the same person across papers is not stated.
Would change if. not applicable (a record that sources differ); the dispute closes only if the early line list is released
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence moderate · what a source says
China CDC reports that after the Huanan market closed on 1 January 2020, 74 of 923 environmental samples (70 by RT-qPCR and 4 by sequencing) were positive for SARS-CoV-2, and none of 457 animal samples (188 animals, 18 species, collected from 18 January) was positive.
Anchor. Liu et al., Nature (read, publisher HTML): positivity 8.0 percent overall; 87.5 percent (56 of 64) of in-market positives in the West Zone; positives in 21 of 134 sampled shops; live virus isolated from three samples with Ct below 30; origin "cannot be determined from the analyses available"; human introduction to the market or the cold chain "cannot yet be ruled out". Note added in proof: the preview's "73" became 70 plus 4.
Would change if. an independent re-analysis of the raw sequencing data changes the positive count
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
Disputed by.
established confidence moderate · what a source says
Published analyses state that no known sarbecovirus other than SARS-CoV-2 has a furin cleavage site at the S1/S2 junction of its spike. In experiments the site matters: removing it weakens SARS-CoV-2 in human airway cells, hamsters, mice and ferret transmission. Furin cleavage sites do occur in other coronavirus groups, and bats are heavily under-sampled.
Anchor. Chan and Zhan 2022 (read): "no known sarbecovirus except SARS-CoV-2 has an FCS insertion at the S1/S2 junction"; they also say "an S1/S2 FCS emerging in a sarbecovirus is consistent with natural evolution". Temmam et al. 2022: "None of these bat viruses contains a furin cleavage site in the spike protein." Andersen 2020: polybasic sites "have not been observed in related 'lineage B' betacoronaviruses, although other human betacoronaviruses, including HKU1 (lineage A), have those sites". Johnson 2021, Peacock 2021 and Hoffmann 2020 show the site's role in experiments. Wu and Zhao 2021: furin sites "occurred independently for multiple times in the evolution of the coronavirus family". SAGO: elements "similar but not identical" are common in several other coronavirus groups.
Would change if. a sarbecovirus with a natural S1/S2 furin site is found in a pre-2019 sample, or a construct carrying it is found in a laboratory
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence moderate · what a source says
RaTG13, the bat virus then closest to SARS-CoV-2, is 96.2 percent identical to it (Zhou et al., Nature, Feb 2020). GenBank records it as submitted by WIV on 27 January 2020 with a collection date of 24 July 2013 and released on 24 March 2020, and replaced by a second version on 13 October 2020. The Nature paper gives no collection year, and WIV's November 2020 addendum discusses sera from four 2012 miners without linking them to RaTG13.
Anchor. GenBank MN996532.1 (read): "Submitted (27-JAN-2020)", "/collection_date="24-Jul-2013"", "On Oct 13, 2020 this sequence was replaced by MN996532.2." Zhou et al. 2020: "overall genome sequence identity of 96.2%" and RaTG13 "previously detected in Rhinolophus affinis from Yunnan province" [read]. The addendum (read) has 13 serum samples from 4 patients received 1 July to 1 October 2012 and confirmed not infected by SARS-CoV-2.
Would change if. WIV releases the sample records and sequencing dates for RaTG13
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence moderate · what a source says
The US Intelligence Community's June 2023 report on the Wuhan Institute of Virology states that, as of January 2019, WIV researchers performed SARS-like coronavirus experiments in BSL-2 laboratories; that some researchers probably did not always use adequate biosafety precautions; that it has no indication WIV's pre-pandemic holdings included SARS-CoV-2 or a close progenitor, no direct evidence of a specific research-related incident and no knowledge of a specific biosafety incident that spurred the pandemic; and that WIV scientists have genetically engineered coronaviruses with common laboratory practices, with no information that this involved SARS-CoV-2, a close progenitor or a backbone close enough to be the source.
Anchor. ODNI report declassified 23 June 2023 (read): "We continue to have no indication that the WIV's pre-pandemic research holdings included SARS-CoV-2 or a close progenitor, nor any direct evidence that a specific research-related incident occurred involving WIV personnel before the pandemic that could have caused the COVID pandemic." "We do not know of a specific biosafety incident at the WIV that spurred the pandemic." "Some WIV researchers probably did not use adequate biosafety precautions at least some of the time prior to the pandemic in handling SARS-like coronaviruses". The report also says it "does not address the merits of the two most likely pandemic origins hypotheses" and that a classified annex is "consistent with the unclassified assessments".
Would change if. the classified annex or the underlying reporting is released, or WIV records are reviewed
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence moderate · what a source says
The 2018 DEFUSE proposal (EcoHealth Alliance to DARPA, requesting USD 14,209,245, with WIV and UNC Chapel Hill as partners) says that, where spike sequences lack cleavage sites, "we will introduce appropriate human-specific cleavage sites and evaluate growth potential in Vero cells and HAE cultures". The public text names no S1/S2 position, gives no sequence, contains no SARS-CoV-2 sequence, and says the chimeric backbone work would be BSL-3.
Anchor. DEFUSE main volume (read, OCR): "We will analyze all SARSr-CoV S gene sequences for appropriately conserved proteolytic cleavage sites in S2 and for the presence of potential furin cleavage sites. ... Where clear mismatches occur, we will introduce appropriate human-specific cleavage sites and evaluate growth potential in Vero cells and HAE cultures." Synthesised spike genes "designed for insertion into SHC014 or WIV16 molecular clone backbones ... These are BSL-3". Cover letter dated 24 March 2018; form field "Date proposal submitted: 3/27/18". Budget line: subcontract to Prof. Baric "to oversee reverse engineering of SARSr-CoVs".
Would change if. the DARPA or EcoHealth holder copy differs from the third-party copy
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence moderate · what a source says
In a 2026 Senate interview Dr Ralph Baric said the DEFUSE furin plan was "loss-of-function first" (remove an existing site, then consider putting one into a null backbone), that "we've never dropped a furin cleavage site into a coronavirus like that" before the application, that one cannot "ever conclusively say that there was not a laboratory leak", and that he was "100 percent certain" he was not involved in creating the virus.
Anchor. Senate HSGAC transcribed interview, 10 April 2026 (read at the passages on the furin plan and BSL-2). He also said that, as written, "That part would be done by me" and that whether it would have happened "is a fair question, because if the Chinese discovered the virus ... whether they would share it with me."
Would change if. laboratory records of the work he describes are produced
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence moderate · what a source says
Official US statements on sick WIV researchers narrowed over time. The State Department (15 Jan 2021) said it had "reason to believe that several researchers inside the WIV became sick in autumn 2019 ... with symptoms consistent with both COVID-19 and common seasonal illnesses". ODNI 2021 called this "not diagnostic". ODNI 2023 said "We have no indications that any of these researchers were hospitalized because of the symptoms consistent with COVID-19", that some symptoms were not consistent with COVID-19, and that the information "neither supports nor refutes either hypothesis".
Anchor. State Department fact sheet, 15 Jan 2021 (Wayback copy); ODNI Updated Assessment ("not diagnostic of the pandemic's origins. Even if confirmed, hospital admission alone would not be diagnostic of COVID-19 infection"); ODNI June 2023 ("One researcher may have been hospitalized in this timeframe for treatment of a non-respiratory medical condition"). WIV side: the WHO-China report records, from lab representatives, a staff health monitoring programme "with no reporting of COVID-19 compatible respiratory illness" and no serological evidence of infection; SAGO says no occupational health records, audits or absenteeism reports were provided [read].
Would change if. WIV occupational-health records or stored sera are independently reviewed
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
HHS suspended and proposed to debar EcoHealth Alliance on 15 May 2024 and debarred it for five years in January 2025 (ineligible through 14 May 2029); it suspended Dr Daszak on 21 May 2024 and debarred him for five years (ineligible through 20 May 2029). The grounds are grant reporting and responsibility failures, not a finding about where the virus came from.
Anchor. Notice of suspension 15 May 2024; final notices and memoranda dated by the House release as 17 Jan 2025 ("EHA will remain ineligible through May 14, 2029"; Daszak "ineligible through May 20, 2029"). The EHA memorandum summarises a WIV suspension on 17 Jul 2023 and a ten-year WIV debarment on 19 Sep 2023; the WIV notice itself was not read. The public documents name private attorneys; they are not named here [read].
Would change if. not applicable (a record of an action)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
The March 2020 Nature Medicine correspondence "The proximal origin of SARS-CoV-2" says "Our analyses clearly show that SARS-CoV-2 is not a laboratory construct or a purposefully manipulated virus", that "we do not believe that any type of laboratory-based scenario is plausible", and in the same section that "it is currently impossible to prove or disprove the other theories of its origin described here".
Anchor. Andersen et al. 2020 (read in full): the two features argued are the receptor-binding domain ("not ideal" and different from computationally predicted optimal) and the polybasic site (no known backbone). The paper's own authors' conditions for revising the conclusion include sequences of related viruses from animals and an intermediate host; Andersen's 2023 written testimony says "If convincing new evidence were to be discovered that suggests the virus emerged from a lab, we would, of course, revise our conclusions."
Would change if. not applicable (a record of what the paper says)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
In 2021 the US Intelligence Community said that four elements and the National Intelligence Council assess with low confidence that natural exposure was most likely, one element assesses with moderate confidence that a laboratory incident was most likely, and three elements could not decide; most agencies assess with low confidence that the virus probably was not genetically engineered, and two could not say. The stated basis is "all available intelligence reporting and other information".
Anchor. ODNI Unclassified Summary (print date 27 Aug 2021) and Updated Assessment (information through August 2021): "Four IC elements and the National Intelligence Council assess with low confidence that the initial SARS-CoV-2 infection was most likely caused by natural exposure to an animal infected with it or a close progenitor virus"; "One IC element assesses with moderate confidence that the first human infection with SARS-CoV-2 most likely was the result of a laboratory-associated incident"; "Analysts at three IC elements remain unable to coalesce around either explanation without additional information"; "The IC—and the global scientific community—lacks clinical samples or a complete understanding of epidemiological data from the earliest COVID-19 cases." The documents do not publish the underlying intelligence [read].
Would change if. not applicable (a record); the agencies' positions changed in 2023 (co-odni-2023-agencies)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
In June 2023 the US report said that the National Intelligence Council and four other agencies assess that natural exposure was most likely, that the Department of Energy and the FBI assess that a laboratory-associated incident was most likely "although for different reasons", and that the CIA and one other agency remain unable to determine. The confidence level of each position is redacted in the public text.
Anchor. ODNI June 2023 (read): "The Central Intelligence Agency and another agency remain unable to determine the precise origin of the COVID-19 pandemic, as both hypotheses rely on significant assumptions or face challenges with conflicting reporting." "Almost all IC agencies assess that SARS-CoV-2 was not genetically engineered. Most agencies assess that SARS-CoV-2 was not laboratory-adapted; some are unable to make a determination."
Would change if. not applicable (a record); the 2025 CIA statement changed one position (co-cia-2025)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
The joint WHO-China report (2021) rated on an "arbitrary Likert opinion scale" direct zoonotic spillover "possible to likely", spillover through an intermediate host "likely to very likely", cold or food chain "possible", and a laboratory incident "extremely unlikely". On the day the report was presented, WHO's Director-General said of the laboratory assessment "I do not believe that this assessment was extensive enough."
Anchor. The report's lab-incident conclusion rests on statements from the laboratories' representatives (health monitoring with no reports of compatible illness; "no record of viruses closely related to SARS-CoV-2 in any laboratory before December 2019"). The Director-General, 30 Mar 2021: "Although the team has concluded that a laboratory leak is the least likely hypothesis, this requires further investigation"; the team had "difficulties ... accessing raw data"; "all hypotheses remain on the table". SAGO 2025 notes verbal assurances rather than physical inspections [read].
Would change if. not applicable (a record)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
WHO's advisory group SAGO (27 June 2025) said "The weight of available evidence reviewed by SAGO suggests zoonotic spillover of SARS-CoV-2 into the human population", could not say where or when, and said information to assess a laboratory origin was "either ... not available or has not been provided", so it "is not in a position to rule this out". It says it had no raw data and that requested information from China, Germany and the United States was not provided.
Anchor. SAGO 2025 (read): "It is important to note that SAGO did not have access to original raw data from any source"; "no evidence has been presented, other than speculation from scientific or intelligence reports, that supports a laboratory-related incident causing the spread of SARS-CoV-2 into the human population. Without the requested information, this hypothesis can however not be ruled out." On the US reports: neither "presented any new or conclusive evidence to support either zoonotic or laboratory origins". On the House majority: claims "without providing any new evidence or factual scientific evidence".
Would change if. not applicable (a record)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
The House Select Subcommittee majority's final report (4 December 2024) finds that SARS-CoV-2 "likely emerged because of a laboratory or research related accident" and that "the weight of the evidence increasingly supports the lab leak hypothesis"; its Democrats' report says "Arguments for a lab origin are largely circumstantial but cannot be dismissed out of hand" and, on whether DEFUSE and SARS-CoV-2 are linked, that "it is unfortunately impossible to draw any conclusions without reviewing additional lab records from WIV".
Anchor. The majority's origin pages (1-5) rely mainly on quoted officials, the DOE, FBI and ODNI assessments and several opinion essays (including one by a journalist and one by a scientist who publicly favours the laboratory hypothesis) for the claims that the virus has a feature "not found in nature" and that all cases stem from a single introduction; these two claims are scientific points sourced to opinion pieces, not to the underlying papers (see co-fcs-not-found-in-nature and co-two-lineages). The report concludes DEFUSE "could have created and released a virus like COVID-19" and that Dr Daszak "initially hid the most dangerous aspect"; the support for the second is a program manager's recollection. The Democrats' report has sections headed "EcoHealth Alliance Did Not Create COVID-19" and "Dr. Fauci Did Not Create COVID-19", and says the Proximal Origin authors confirmed Drs. Fauci and Collins did not lead, oversee or influence the drafting; the majority disputes these [read].
Would change if. not applicable (a record)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
Senate HELP minority staff (October 2022) concluded that the pandemic was "more likely than not, the result of a research-related incident", while saying "This conclusion is not intended to be dispositive"; the American Society for Microbiology replied that it does not believe there is enough available evidence to support the conclusion in the report. A 2023 executive summary by Sen. Marshall's office says "Both hypotheses are plausible".
Anchor. HELP staff (read): "the hypothesis of a natural zoonotic origin no longer deserves the benefit of the doubt"; also that precedent "favors the hypothesis that a natural zoonotic spillover" and that PRC non-transparency "prevents reaching a more definitive conclusion". Marshall 2023 (read, executive summary only): "The preponderance of circumstantial evidence supports an unintentional research-related incident." ASM (28 Oct 2022): "ASM does not believe there is enough available evidence about the origins of SARS-CoV-2 to support the report's overarching conclusion". Weaknesses visible in the documents: the database claim rests on one volunteer-group report (co-sq-wiv-database); the DEFUSE location claim is unsupported by the public text (co-help-defuse-location-claim); one footnote cites a paper that does not match its claim.
Would change if. not applicable (a record)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
China's State Council Information Office white paper (April 2025) says "It is fair to say that the study on the origins of SARS-CoV-2 conducted in China has ended", that the virus may have entered China by cold-chain transport, and that a section of "Evidence Pointing to the US as the Origin of Covid-19" shows the US needs to be investigated. It cites no new evidence for the US claims beyond US sources it names.
Anchor. White paper (read): "To date, no findings have contradicted the conclusions of the 'Joint WHO-China Study'"; "there is a possibility that the outbreak in the market at the end of 2019 was introduced to China from abroad via cold-chain transportation"; it says "all 457 animal samples ... tested negative ... while 74 out of 923 environmental samples from the market were positive"; and that three US entities (Los Alamos, NIH, ODNI) concluded that WIV had "strict biosecurity protocols". The ODNI text read says some WIV researchers "probably did not use adequate biosafety precautions at least some of the time"; the Los Alamos and NIH reports were not read. SAGO notes the paper gives no new evidence for SAGO to review.
Would change if. not applicable (a record)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · what a source says
A White House web page (published 18 April 2025, modified 15 August 2025) states "A lab-related incident involving gain-of-function research is the most likely origin of COVID-19" and lists five arguments, citing no evidence on the page and drawing its content from the House majority report.
Anchor. Page (read) item 4: "Wuhan Institute of Virology (WIV) researchers were sick with COVID-like symptoms in the fall of 2019"; the US report says the symptoms were "consistent with but not diagnostic of COVID-19" (co-wiv-illness-official-record). Item 1 ("a biological characteristic that is not found in nature") is addressed at co-fcs-not-found-in-nature, item 2 ("all COVID-19 cases stem from a single introduction") at co-two-lineages.
Would change if. not applicable (a record)
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
contested confidence low · our conclusion
SARS-CoV-2 emerged through natural spillover from animals, either directly from bats or through an intermediate host, most likely through the live-animal trade in Wuhan.
Anchor. Worobey 2022 (as published), Liu 2023, Crits-Christoph 2024, Bloom 2023, Pekar 2022 and the WHO-China report 2021, all read; see co-market-pattern, co-market-samples, co-wildlife-dna-co-occurrence, co-two-lineages and co-who-china-early-cases. SAGO: "it is not conclusive that the HSM was where the virus first spilled over."
Would change if. an infected animal or a progenitor virus is found in the market's supply chain, or a bias-free case series shows the December clustering came from how cases were found
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
Disputed by.
contested confidence low · our conclusion
As published in 2022, the earliest December 2019 cases, including those with no known market link, were geographically centred on the Huanan market in a way that case-finding does not explain.
Anchor. Worobey et al. 2022 (pre-erratum text, read): coordinates for 155 of 164 Wuhan-resident December cases digitised from the WHO-China report's maps; the market lies inside the highest-density 1 percent contour; median distance to the market 4.28 km for all cases against a null median of 16.11 km; 4.00 km for cases without a known market link and 5.74 km for linked cases (unlinked closer, p=0.029). Reply: Debarre and Worobey 2024 (arXiv, read) say linked cases are mainly workers and unlinked cases lying closer is expected. The published JRSSA versions of the two critiques and the Science erratum were not read.
Would change if. the case line list with onset dates and coordinates shows market-linked case-finding drove the clustering, or the corrected 2024 version of the paper changes its central result
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
Disputed by.
contested confidence low · our conclusion
In market samples from one wildlife stall, SARS-CoV-2 RNA was found together with DNA of raccoon dog, civet, bamboo rat and other mammals, which makes those species candidate intermediate hosts.
Anchor. Crits-Christoph et al. (NIH author manuscript, read): five SARS-CoV-2-positive stall-A samples contained mitochondrial DNA of raccoon dog, hoary bamboo rat, dog and European rabbit; all five also contained human DNA; no bat or pangolin material. Authors' limits: cannot say which species shed the virus. Bloom 2023 (read): only 21 of 176 samples have more than ten SARS-CoV-2 reads.
Would change if. serology or virus sequencing from the live animals or their farms, or unbiased re-sequencing of all January 2020 samples
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
Disputed by.
contested confidence low · our conclusion
Early SARS-CoV-2 genomes fall into two lineages (A and B) whose pattern implies at least two separate cross-species transmissions into humans.
Anchor. Pekar et al. 2022 as originally published (read): 787 lineage A/B genomes to 14 Feb 2020; Bayes factor 61.6 (recCA rooting) and 60.0 (unconstrained); lineage B introduction about 18 Nov 2019 (23 Oct-8 Dec). The authors list limits: possible lineage-assignment errors and no direct evidence of a related virus in market animals. The erratum (Science 382:eadl0585) was NOT read.
Would change if. the Pekar authors' corrected analysis or an independent re-analysis is read and the support for two introductions rises or falls
Why this confidence. Starts at moderate ().
These reasons were drafted by the project and have not yet been reviewed.
Disputed by.
contested confidence low · our conclusion
The first human infection with SARS-CoV-2 was the result of a laboratory-associated incident (experimentation, animal handling or sampling) connected with the Wuhan Institute of Virology.
Anchor. ODNI June 2023 (read): "no indication that the WIV's pre-pandemic research holdings included SARS-CoV-2 or a close progenitor, nor any direct evidence that a specific research-related incident occurred"; "Some WIV researchers probably did not use adequate biosafety precautions at least some of the time"; as of January 2019 WIV researchers performed SARS-like coronavirus experiments in BSL-2 laboratories. DEFUSE proposal text (read): co-defuse-text-fcs-plan. Furin site: co-fcs-unique-among-sarbecoviruses.
Would change if. a laboratory record, sample or sequence shows SARS-CoV-2 or a progenitor was held there, or an infected laboratory worker is documented, or a natural progenitor and host are found
Why this confidence. Starts at high ().
These reasons were drafted by the project and have not yet been reviewed.
Disputed by.
contested confidence low · our conclusion
The Proximal Origin paper was prompted by Dr Anthony Fauci in order to disprove the laboratory hypothesis.
Anchor. Andersen's written testimony (read): the paper was posted 17 Feb 2020 and published 17 Mar 2020, 45 days after the 1 February call; the single suggested edit (from Dr Farrar) changed "unlikely" to "improbable". He also quotes his own 12 Feb 2020 email to Nature ("Prompted by Jeremy Farrah [sic], Tony Fauci, and Francis Collins ...") and recalls Dr Fauci telling him on 31 Jan 2020 to write a paper if he thought the virus came from a lab; the call of 1 Feb 2020 was, he says, organized and convened by Dr Farrar. The hearing transcript (read) holds members' quotations of a 31 January 2020 email and the "crackpot theory" email; the original messages were not read (co-sq-proximal-origin-records).
Would change if. the original January to February 2020 messages and emails are read and show who asked for the paper and what changed
Why this confidence. Starts at moderate ().
These reasons were drafted by the project and have not yet been reviewed.
Disputed by.
proposed confidence provisional · what a source says
The October 2023 Science erratum to Pekar et al. reduced the Bayes factor for two introductions against one from about 60 to about 4 (McCowan) or about 4.3 (Weissman).
Anchor. Two critics, both disputing the paper, describe the erratum: McCowan (arXiv v3, read) "the Bayes factor dropped to 4" after three coding errors; Weissman (EJW 2026, read) "a less compelling ~4.3". They disagree in the second digit; neither is the erratum. PubMed and Crossref confirm that an erratum exists, not its content.
Would change if. the erratum text, once read, gives a different figure
Next step. Read the erratum (Science 382:eadl0585) and the corrected online paper; see co-sq-science-errata.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
proposed confidence provisional · what a source says
The DEFUSE proposal says it would insert furin cleavage sites "at the same location where one appears in SARS-CoV-2" (Senate HELP minority staff, 2022).
Anchor. The HELP report (read, p.15) states it. The location statement was searched for in three copies of the public proposal (the OCR text, a second OCR copy and the PDF text) and not found; the report's footnotes cite a copy "on file with staff", which may be a different version. Absent from the public text; not shown absent from every version.
Would change if. the staff copy of the proposal that the report cites ("on file with staff") is shown to contain it
proposed confidence provisional · what a source says
Draft DEFUSE documents released by FOIA contain a draft sentence that "The BSL-2 nature of work on SARSr-CoVs makes our system highly cost-effective", and a comment tagged [BRS17] that in the US these recombinant viruses are studied at BSL3 and that "In china, might be growin these virus under bs12. US reseachers will likely freak out."
Anchor. USGS FOIA release 2021-006245, OCR text read at the passages quoted (draft email of 8 Feb 2018): the [BRS17] tag's initials are consistent with Ralph S. Baric but the OCR does not state the commenter's name, so the identity is inferred. The uploader says the file was not vetted. The spelling is the OCR text's.
Would change if. the USGS FOIA release itself is compared and the text or tags differ
Next step. Compare with the USGS release or the first pages posted by the news organisation that received them.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
proposed confidence provisional · what a source says
EcoHealth Alliance carried out the DEFUSE proposal until April 2020 through NIH funding, as alleged in a 2022 Senate letter relaying a disclosure by a military officer.
Anchor. Sen. Ron Johnson's letter to the Department of Defense of 11 Jan 2022 (OCR text, read) says "the disclosure alleges" this. No document read supports it. The HHS records document work under a different grant (R01AI110964). SAGO and the House Democrats say DEFUSE was not funded. The officer is not named here.
Would change if. documents are produced that show the DEFUSE work, as opposed to the separately funded NIH grant, was done
proposed confidence provisional · our conclusion
SARS-CoV-2 was genetically engineered in a laboratory.
Anchor. Andersen 2020 (read): "Our analyses clearly show that SARS-CoV-2 is not a laboratory construct" (see co-proximal-origin-text; criticised as stated too strongly). Chan and Zhan 2022 (read): a workflow existed that would make a recombinant virus "practically indistinguishable from a rare SARSr-CoV with a naturally emerging FCS". Neither side has the laboratory records that would settle it.
Would change if. a construct, record or sequence of work that produced the virus is found
proposed confidence provisional · our conclusion
WIV researchers were infected with COVID-19 in autumn 2019, before the first identified case.
Anchor. No test result or diagnosis is public. ODNI 2023: symptoms "consistent with but not diagnostic of COVID-19", "a range of symptoms consistent with colds or allergies", and "some of them were confirmed to have been sick with other illnesses unrelated to COVID-19". China's national security commission is said by ODNI to have taken blood samples in early 2020 that WIV officials say were negative for antibodies.
Would change if. sera, health records or test results from the researchers are independently reviewed
proposed confidence provisional · what a source says
A WIV scientist said in a December 2024 conference talk that WIV holds whole genomes of 56 new betacoronaviruses collected 2004 to 2021, and that none is more closely related to SARS-CoV-1 or SARS-CoV-2 than those already known.
Anchor. Nature news reposted by Scientific American, 6 Dec 2024 (read): "We didn't find any new sequences which are more closely related to SARS-CoV-1 and SARS-CoV-2"; the analysis "has not been peer reviewed". A limited search of Europe PMC found no later peer-reviewed paper with the data. The talk itself was not seen.
Would change if. the sequences are published and independently checked
Next step. Obtain the sequence set and the collection records for independent analysis (dt-wiv-records).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
proposed confidence provisional · what a source says
Private Slack messages among the Proximal Origin authors, released by the Senate committee chairman in 2026, show on 11 June 2020 one author writing that the paper "was pretty strong in saying" there was no way, but "I have less confidence in that statement at this stage", and the same author writing later that day "still believe this is a natural virus".
Anchor. HSGAC release of 22 July 2026, volume 1 (read, SLACK_000283-287): "we really don't have any hard evidence one way or the other ... we also can't rule out that somebody actually put it in there. That's obviously not to say that somebody"; "given no data, I find both scenarios to be equally probable"; another author writes "To me, there is more evidence against lab escape than when we wrote the paper". Numbers exchanged later that day ("80/20", "70/30", "90/10") do not state which hypothesis each refers to; the press release reads them as lab-origin odds, which is a reading. Unchecked source: no independent capture; the file is chairman-selected and does not cover January and February 2020.
Would change if. the originals are compared with the released set and differ
Next step. Compare with originals released elsewhere or the authors' own copies.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
proposed confidence provisional · what a source says
The FBI assesses that a laboratory-associated incident was the most likely cause (named in the US June 2023 report; its Director said in February 2023 that "the FBI has for quite some time now assessed that the origins of the pandemic are most likely a potential lab incident in Wuhan"). Its "moderate confidence" is reported only by press; its basis is unpublished.
Anchor. ODNI 2023 (read) names the FBI. The Director's words are as quoted by CNN of 28 Feb 2023 (read), which adds that CNN reported in 2021, citing two sources familiar with the matter, "moderate confidence". fbi.gov refused this environment; see co-sq-agency-documents.
Would change if. an FBI document with its confidence and basis is published
proposed confidence provisional · what a source says
The Department of Energy assesses that a laboratory-associated incident was the most likely cause (named in the US June 2023 report). Its "low confidence" is reported by NPR relaying the Wall Street Journal; the report and its reasons are classified.
Anchor. ODNI 2023 (read) names DOE. NPR of 28 Feb 2023 (read): "the intelligence in it is classified"; DOE "with low confidence". The Wall Street Journal article (26 Feb 2023 per NPR) was not read.
Would change if. a DOE document with its confidence and basis is published
proposed confidence provisional · what a source says
A CIA spokesperson was quoted in January 2025 as saying "CIA assesses with low confidence that a research-related origin of the COVID-19 pandemic is more likely than a natural origin based on the available body of reporting", and that both scenarios "remain plausible". CBS and AP say the finding was not the result of new intelligence.
Anchor. CBS News with AP (page published 26 Jan 2025, updated 27 Jan 2025; read): released Saturday 25 Jan 2025 on the orders of the new Director; "no new 'smoking gun'" according to two officials. About ten cia.gov addresses tried in the research stream returned 404. In the 2023 US report the CIA was "unable to determine"; SAGO (June 2025) says a CIA report "is not publicly available and thus has not been reviewed".
Would change if. the CIA's own release is read and differs from the quotation
Next step. Locate the CIA's declassified document (see co-sq-agency-documents).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
proposed confidence provisional · what a source says
German press reported in March 2025 that the German foreign intelligence service (BND) had judged a laboratory accident probable since 2020, with a probability reported as 80 to 95 percent (Tagesspiegel and NOS) or 80 to 90 percent (as put by a journalist at a government press conference). The German government declined to comment.
Anchor. Tagesspiegel, 12 Mar 2025 (lead only, read): "80 bis 95 Prozent"; NOS 15 Mar 2025 (read) relays it, with a project name and unpublished Chinese documents, and "Onomstotelijk bewijs is er niet". Government press conference, 12 Mar 2025 (read): the journalist said "80 bis 90 Prozent"; the spokesperson said the government cannot comment on intelligence activities. The Bundestag question of 21 Mar 2025 (read) recounts the report; no reply was found.
Would change if. the BND assessment is released or confirmed
Next step. Read the Zeit and Süddeutsche originals and any government reply to Bundestag question 20/15143.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
The origin of SARS-CoV-2 has not been established: no document or sample read for this dig shows an infected animal, an intermediate host, a progenitor virus, or a laboratory that held, made or released the virus.
Anchor. Read in full or in the passages used (2026-10-05): ODNI 2021 and 2023 ("no confirmed animal source of SARS-CoV-2 has been identified" and "no indication that the WIV's pre-pandemic research holdings included SARS-CoV-2 or a close progenitor"), the WHO-China report 2021, SAGO 2025, both House reports, both Senate staff products, the DEFUSE proposal and drafts, and the Baric transcript. None presents an infected animal, intermediate host or progenitor, or a laboratory record of one. This is a search of the documents we could open; the classified annex to the US report, China's laboratory records and its early data were not available to us.
Would change if. a virus more than 99 percent similar to SARS-CoV-2 is found in an animal population or in a laboratory's records or freezers, or an infected market animal or its supplier is identified
Next step. Obtain the items listed in the discriminating tests: early sequences, market-animal and supply-chain sampling data, and laboratory records (dt-progenitor-search, dt-early-sequences, dt-market-animals, dt-wiv-records).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
No virus more than 99 percent similar to SARS-CoV-2 (the figure the US report uses for a close progenitor) has been published from any animal or any laboratory. The closest known relatives are 96.8 percent (BANAL-52) and 96.2 percent (RaTG13) similar.
Anchor. ODNI 2023 (read) gives RaTG13 as 96.2 percent and BANAL-52 as 96.8 percent similar and says "Neither of these viruses is close enough to SARS-CoV-2 to be a direct progenitor", and defines a close progenitor as "a virus that probably would be more than 99 percent similar". Temmam et al. 2022 (read) report the Laos viruses; Zhou et al. 2020 (read) report RaTG13 at 96.2 percent. Searched in the documents listed in co-origin-not-established; no record of a more similar virus was found. Not shown: that none exists in closed records. A Dec 2024 conference talk by a WIV scientist reported no new sequence closer to SARS-CoV-1 or SARS-CoV-2 (co-shi-2024-sequences; secondary).
Would change if. a virus more than 99 percent similar to SARS-CoV-2 is published from any source
Next step. Sample bats, other wildlife and market supply chains in and around Hubei and Yunnan and Laos, and release WIV's sequence archives for independent review (dt-progenitor-search, dt-wiv-records).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
No SARS-CoV-2 test result exists for any of the live mammals sold at the Huanan market in late 2019; the 457 animal samples were taken after closure from carcasses, freezer and refrigerator contents, stray animals and a fish tank, and were all negative.
Anchor. Worobey et al. (read) list live mammals seen at the market in late 2019 and state that no SARS-CoV-2 test results for any of these animals exist; Liu et al. (read) report the 457 samples collected from 18 January 2020. SAGO lists "more detailed information on the sources, locations, sampling and testing methods of animals sold at the HSM" among data not provided. A search of the documents read; absence in them is not proof of none held.
Would change if. test results from the market's live animals or their suppliers are produced
Next step. Release the sources, locations, sampling and testing data for animals sold at the market and wet markets in Wuhan, including upstream farms and trade (dt-market-animals).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
What the two Science errata (Worobey et al., 14 Mar 2024; Pekar et al., 13 Oct 2023) changed has not been read for this dig.
Anchor. Tried 2026-10-05: science.org (403), Europe PMC and NCBI full-text XML (500 or not permitted), Crossref (metadata only, confirming an update of type correction for both). No secondary source for the Worobey erratum was found; the Pekar erratum is described only by two critics (co-pekar-erratum-bf).
Would change if. the errata are read and figures quoted in this dig differ from the corrected versions
Next step. Read both errata and the corrected papers at science.org (refused this environment, HTTP 403; Europe PMC withholds the erratum full text), then re-check co-market-pattern and co-two-lineages.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
Sampling, testing and supply-chain records for the animals sold at the Huanan market and other Wuhan wet markets have not been made available to outside reviewers.
Anchor. SAGO 2025 (read) lists "more detailed information on the sources, locations, sampling and testing methods of animals sold at the HSM and other wet markets in Wuhan, including upstream farming or illegal trade" among information not provided; the WHO-China report recommends supply-chain trace-back with farm sampling. No such data were found in any document read.
Would change if. the records are released
Next step. Release or independently audit the supplier, farm and sampling records (dt-market-animals).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
DARPA's own decision record on DEFUSE (the letter, the grading and its reasons) has not been read; the sources differ on why it was not funded.
Anchor. The House report (read, pp.116-119) says the proposal was graded "selectable, but not recommended" and that the program manager said the lack of a gain-of-function plan affected the decision; Daszak (as quoted there) says biosafety "never once did ... come up" and that it was too much money; Baric (2026 transcript, read) says the reason given to him concerned in-cave intervention studies. A file named as a DARPA rejection letter in the research set was not a DARPA letter (it is a summary by an advocacy group) and was not used.
Would change if. the DARPA letter or evaluation record is released
Next step. Obtain the DARPA letter that the House report cites as "on file with staff" and the program manager's evaluation record.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
Whether WIV's online Wildlife-Borne Viral Pathogen Database was taken offline on 12 September 2019, and what it held, has not been verified for this dig.
Anchor. Claim as written (HELP minority staff 2022, read): the database was taken down on 12 September 2019 "between the hours of 2:00 and 3:00 a.m. local time" and held more than 15,000 samples and more than 1,400 bat viruses; staff attempts to reach it on 18 Oct 2022 "were unsuccessful". The Marshall 2023 summary (read) says "approximately 20,000" samples and "more than 2,000 entries". Both cite one volunteer-group report and a newspaper editorial, which were not opened (zenodo and researchgate refused). No archive capture was found. The ODNI, WHO-China and SAGO texts searched contain no statement about this database.
Would change if. dated archive captures of the database, or its contents, are produced, or WIV explains the removal
Next step. Find dated archive captures of batvirus.whiov.ac.cn, and the volunteer-group report the Senate documents cite; ask WIV for its server logs.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
WIV's laboratory records and staff health records have not been made available to outside reviewers.
Anchor. SAGO 2025 (read): WHO wrote to China to request "further investigations, audits and other relevant documentation and interviews with staff at the WIV and Wuhan CDC laboratory"; "this has been refused"; and, on its data requests to Member States, "all of the required information had not been provided to WHO". The HHS record (read): NIH asked in Nov 2021 and Jan 2022 for original notebook entries and electronic files, and "as of October 2024 ... NIH had yet to receive any records". ODNI 2023 says the IC "assesses that the WIV maintains blood samples and health records of all of their laboratory personnel" (an assessment, not a document seen).
Would change if. the records are released or independently audited
Next step. Independent audit of WIV records and stored sera (dt-wiv-records).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
More than 500 sequences from early COVID-19 patients and the early-case line list have not been shared outside China.
Anchor. SAGO 2025 (read) lists among data not provided "the sharing of more than 500 sequences from individuals with COVID-19 early in the pandemic". Worobey et al. (read) say "no line list of early COVID-19 cases is available". Bloom 2021 (read; a 2023 correction was not read) recovered a deleted dataset of 45 early outpatient swabs and reconstructed partial sequences for 13, concluding that "the progenitor of currently known SARS-CoV-2 sequences likely contained three mutations relative to the market viruses"; the paper calls the proximal origin "unclear (i.e., zoonosis versus lab accident)" and concludes that market sequences "are not fully representative"; it does not claim its data decide between the two.
Would change if. the sequences and the line list are released
Next step. Release or audit the early sequences and the line list with onset dates and coordinates (dt-early-sequences).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
The original January and February 2020 messages and emails behind the Proximal Origin paper have not been read for this dig; only quotations in a hearing transcript and a witness's testimony were.
Anchor. The released Slack files read begin in April 2020. The testimony's email images (8 and 12 Feb 2020) were not extracted. Press and social-media postings of the early messages surfaced in search and were not fetched.
Would change if. the original messages are read
Next step. Read the originals as released by the House Select Subcommittee; compare the testimony's email images.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
A March 2025 US court judgment against China (a default judgment of about USD 24.49 billion, per China's white paper) has not been read; the only source read for it is China's white paper, which rejects it.
Anchor. China's white paper (read) says a US court default judgment of 7 March 2025 awarded USD 24.49 billion and calls it a "politically motivated farce". The document itself was not read, so what the judgment does and does not find about origin is not stated here.
Would change if. the judgment is read
Next step. Read the court's judgment and say what it does and does not find about origin.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
The primary documents of the FBI, DOE, CIA and German BND assessments have not been read; each position is known only through the US June 2023 report or news reports.
Anchor. fbi.gov returned 403; no DOE publication was found; about ten cia.gov addresses returned 404; the BND document is not public. Tried 2026-10-05 in the research stream.
Would change if. the documents are published and read
Next step. Obtain the FBI, DOE, CIA and BND documents, or their classified annexes if released.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence provisional · result of a search
The documents the Director of National Intelligence released on 18 June 2026 (four parts and an index) have not been read; only the introduction on ODNI's page was.
Anchor. ODNI's page (read, dated 18 June 2026) says the Director is releasing "never-before-seen communications and documents" and alleges that a named public official influenced IC assessments and misled Congress. These are the page's allegations; no release document was opened, so nothing is stated here about what they show.
Would change if. the release is read; it may bear on how agency assessments were produced
Next step. Read the four release parts and the index and record what they show, separately from the page's framing.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
Publication status is recorded only for sources someone looked up. A source with no status shown has not been checked for a retraction or correction; that is not the same as clean.
These pages are generated from plain YAML: claims.yaml · sources manifest. If you can show a finding is wrong, challenge it.