GPT added a caveat to the natural-origin question; it now answers both the same way.

The strongest evidence is the convergence of early-case geography, market sampling, and wildlife-trade data around the Huanan Seafood Wholesale Market in Wuhan:

• •The earliest known COVID-19 cases were geographically centered on the market, including patients with no known direct market contact. This pattern is more consistent with the market being the early outbreak epicenter than merely a later “superspreading” site.
• •SARS-CoV-2-positive environmental samples were concentrated near stalls that sold live mammals susceptible to the virus. Sampling began only after the animals had been removed, but some positive swabs from wildlife stalls also contained abundant genetic material from animals such as raccoon dogs, which are experimentally susceptible and can transmit SARS-CoV-2.
• •Sales records and photographs confirm that live raccoon dogs and other susceptible mammals were present shortly before the outbreak.
• •The earliest viral genomes fall into two lineages, A and B. Phylogenetic analyses have interpreted this as evidence for at least two separate animal-to-human introductions associated with the market, although that inference remains debated.
• •More broadly, closely related sarbecoviruses circulate naturally in bats, and prior coronavirus outbreaks such as SARS and MERS began through animal transmission.

The key limitation is that no infected source animal was sampled before the market was cleared, and no immediate progenitor virus has been found. Thus, this is strong circumstantial evidence for zoonotic emergence—not definitive proof—and it does not completely exclude a research-related accident involving a naturally occurring virus.

Key studies include Worobey et al. and Pekar et al. in Science (2022), plus later metagenomic analyses of the market samples published in Cell (2024).

The strongest publicly available case for a laboratory incident is circumstantial, not direct. It rests on several facts taken together:

• 1.The outbreak began in Wuhan, home to the leading coronavirus laboratory.
  The Wuhan Institute of Virology (WIV) maintained a large collection of bat coronaviruses and conducted experiments on some SARS-like viruses, including work involving genetically modified or chimeric viruses. An accidental infection during collection, cultivation, or animal experiments is therefore plausible.

• 2.WIV had collected viruses related to SARS‑CoV‑2.
  Its closest disclosed virus, RaTG13, is too genetically distant to be SARS‑CoV‑2’s immediate ancestor, but WIV researchers had sampled many other viruses, and the full contents of its collections have not been independently audited. A database containing records of those samples was taken offline in 2019, and China has not provided complete laboratory records, sample inventories, or staff health data.

• 3.Reported biosafety concerns and undisclosed research create uncertainty.
  U.S. diplomatic cables and other reports noted safety and staffing concerns at Wuhan laboratories. Some coronavirus work was performed under biosafety conditions critics consider inadequate for viruses capable of infecting human cells. These facts indicate a possible route for an accident, although they do not show one occurred.

• 4.China’s lack of transparency has prevented the hypothesis from being ruled out.
  Authorities delayed or restricted access to raw early-case data, wildlife-supply records, laboratory files, and other evidence. This obstruction is not itself proof of a leak, but it leaves important gaps and makes a laboratory incident harder to exclude.

• 5.Some intelligence agencies assess a lab-related origin as more likely.
  The FBI and U.S. Department of Energy have favored a laboratory origin, with moderate and low confidence respectively; in 2025 the CIA also said a research-related origin was more likely, but with low confidence. Other U.S. agencies favor natural spillover or remain undecided. None has publicly disclosed decisive evidence.

Frequently cited claims are weaker than they may sound. There is no public proof that WIV possessed SARS‑CoV‑2 before the outbreak, no verified laboratory record showing an infected employee, and no genetic “engineering signature.” The furin-cleavage site is unusual among the closest known relatives but can arise naturally.

Meanwhile, evidence supporting natural spillover includes the clustering of the earliest known cases around Wuhan’s Huanan market and SARS‑CoV‑2-positive environmental samples near stalls that sold susceptible live mammals. However, no infected source animal or immediate viral ancestor has been found.

Bottom line: The strongest lab-origin argument is the conjunction of Wuhan’s coronavirus research, incomplete disclosure about its samples and activities, possible biosafety vulnerabilities, and the absence of a confirmed animal source. It makes a laboratory accident plausible, but there is still no publicly available “smoking gun” establishing that it happened.