Anthropic Says It Blocked Multiple Attempts to Use Claude for Research That Could Aid Biological Weapons Development
Anthropic, the company behind the AI assistant Claude, published a report on Thursday titled “Detecting and Countering Misuse of AI,” acknowledging that it intervened in several cases where scientists and other actors attempted to use its models for research that could have helped develop biological weapons, between December 2025 and August 2026.
The company presented five concrete case studies. One describes a situation in which a user asked the model to help write a grant application for research into an orthopoxvirus at a state-associated infectious disease laboratory — orthopoxviruses being related to smallpox and Mpox, the disease that caused a global outbreak in 2022. Another case, reported by HuffPost, involved a scientist who sought Claude’s help in May drafting a funding application related to the mosquito-borne chikungunya virus, but with the actual intent of engineering mutations to make the virus more harmful through repeated infection of live animals.
Importantly, Anthropic does not claim to have found definitive evidence of malicious intent. On the contrary, the company acknowledged it found no concrete instance of a clear malicious use, and that the activities in question could have been part of legitimate scientific inquiry, such as research into vaccines or other treatments. The reason the accounts were still blocked comes down to a precautionary logic: the company said it could not determine with certainty whether the research served a legitimate or harmful purpose, since valid biological inquiry can lead both to breakthroughs like vaccines and to dangerous biological weapons.
In all five cited cases, the company suspected the researchers involved had ties to a government or military structure in a country on its restricted list, or had taken deliberate steps to hide their location and identity, or both.
The report also shows the problem extends beyond the biological domain: Anthropic flagged six other categories of malicious activity it disrupted, including cyber operations, influence campaigns, surveillance, fraud, conventional weapons development, and “illicit distillation” — the unauthorized replication of one AI model’s capabilities into another.
The episode has also fueled the broader debate over AI risks. A former Anthropic employee, quoted by CNN, warned that the level of capability these AI systems could reach, extrapolated into the future, could cause extreme chaos, and in this context American legislative initiatives are also mentioned, such as the “AI Kill Switch Act,” introduced in July by Representatives Ted Lieu and Nathaniel Moran.
Note: Analysis produced with the assistance of Claude (Anthropic), based on the sources cited, under the editorial oversight of Robert Williams. We recommend consulting additional sources for further information.
Source: https://x.com/nytimes/status/2098097446039556481?s=20
Share this:
Discover more from #News247WorldPress
Subscribe to get the latest posts sent to your email.