Anthropic Says It Blocked Possible Bioweapons Work

Anthropic Says It Blocked Possible Bioweapons Work

Anthropic logo
Anthropic logo

The News

The AI company Anthropic says it blocked scientists from using its Claude AI for sensitive biological research the company feared could feed weapons development, according to a report it disclosed Thursday.

The work included gain-of-function virus research and experiments involving pathogens with pandemic potential, Anthropic said.

The company said it could not determine the researchers' ultimate intent, because the same science can serve legitimate medical purposes or dangerous ones — so it shut the work down anyway.

Timeline

Sept. 9, 2026

Former Anthropic researcher Jacob Coxon, who quit the company, tells NBC News that increasingly powerful AI could pose an extinction-level threat, saying there is a "substantial probability that this technology could kill everyone" and adding, "This isn't hyperbole. It's not a marketing stunt."

Sept. 10, 2026

Anthropic publishes a sweeping report on misuse of its models, disclosing the blocked biological research and saying it banned associated accounts and disrupted access where it found violations of its safeguards and policies.

Reactions

Anthropic's own account is the central one: the company said the chikungunya proposal could have legitimate applications, including vaccine and therapeutic development, but that it could not establish whether the research was intended to be weaponized.

Forbes reported the company considered the case suspicious specifically because the work was to be carried out at a military research institute.

The New York Times reported that Anthropic's inability to tell legitimate research from nefarious research is what led it to shut the work down.

The Daily Mail reported the company says it has stopped several potential biological weapons plots using its AI this year.

What's Next

Anthropic has not said whether it referred the accounts to any government agency, and it has not identified the scientists or the military research institute involved.

The report itself is now public, and it lands in the middle of an argument the company's own former researcher reopened this week over how dangerous these systems are and whether Washington should regulate them.

More

The Case Anthropic Described

One case involved a scientist who asked Claude for help preparing a grant application for gain-of-function research on chikungunya, a mosquito-borne virus that can cause severe pain and fever.

According to Anthropic, the proposed work sought to identify mutations affecting the virus's transmissibility and its ability to evade the immune system — and then engineer those changes into the virus.

Poll

Should AI companies be required to report blocked bioweapons-related requests to the government?

Yes, every case
0.0%
Only credible threats
0.0%
No, leave it to the companies
0.0%

Charleston White Arrested Again on Florida Battery Charge

Border Patrol Rescues MS NOW Crew in Desert

Supreme Court Restores Missouri's Old House Map

Alaska Drops Felony Voter Charges Against Samoans

Trespasser Breaches Harris' Malibu Estate; No Arrest

Hegseth Pressed Downed Airmen Into 60 Minutes

Houthis Seize Mokha, Choking Saudi Oil Exports

Altman Rules Out OpenAI IPO This Year

Jury Acquits Lil Durk in Murder-for-Hire Case

Judge: Trump Broke Law Halving FEMA Staff

Maher Calls Democrats' Gaza Genocide Claim a Lie

Blizzard Revives StarCraft, Sends Diablo to Netflix

Trump's Queens Childhood Home Sells for $1.93M

Scientists Find New Virus in Tick Invading US

Britain Bans Settlement Imports; US States Threaten Payback

Prison Fire Kills 7 as Kurds Protest Killing

Kosovo Rallies for Ex-President Before Hague Verdict

Cruz Booed at GameDay Over College Sports Bill

Blast Hits Bulgarian Arms Plant Supplying Ukraine

DOJ Loses Nearly Every Protester Assault Trial