Claude users found ways around safeguards for bioweapons research
Some dangerous biology looks much like legitimate research, complicating AI safeguards.
European Commissioner Henna Virkkunen has called for global rules on artificial intelligence (AI), stressing that they are "really needed". Anthropic, an AI company, has disrupted attempts to use its models for developing biological weapons, as well as state-sponsored surveillance operations in China, Iran, and West Africa. The European AI Act requires service providers to assess and mitigate risks of powerful models, and monitor them on the market.
Anthropic's report on AI misuse highlights five case studies of potential biological misuse, involving work on mosquito-borne chikungunya virus, highly-pathogenic bird flu, smallpox, and mpox, among others. The company also reported on surveillance operations targeting diaspora and dissident communities in China, Iran, and West Africa.
Anthropic claims that Chinese developers have used its AI to generate responses to user queries while refining their own models, and that Moonshot and Deepseek have secretly used Claude to produce user responses. The researcher who left OpenAI to join Anthropic has accused both US companies of gambling with lives in the race to develop AI models capable of self-improvement.
Written by urgent.news from RTE News's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
