Anthropic Blocked 5 Misuses of Claude for Bioweapons — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic disclosed Thursday it disrupted five potential instances of actors using its models to support biological weapons development, detailing real-world case studies uncovered between December and August.
- In two of the five cases, researchers attempted to use Claude for assistance with gain-of-function research on dangerous viruses, per the company's threat assessment.
- Stanford researchers reported last month they used generative AI to design a synthetic virus — the first time the technology was harnessed to create an organism not seen in nature.
- A survey of more than 100 national security experts by the Institute for Security and Technology found 70% believe AI meaningfully increases bioweapon risk or will within two to three years.
- Researchers from Fordham, Johns Hopkins, Oxford, Stanford, Columbia and NYU warned current AI models can design new virus shells, forecast pathogen evolution, create nucleic acid sequences that evade safety-screening software, and synthesize more powerful viral genomes.
- AI developers are releasing new, more efficient biological models often without conducting basic safety assessments — a practice that "wouldn't be tolerated in other parts of life science research," the researchers wrote.
- A "kill switch" bill in Congress would give the government authority to deactivate AI models capable of causing catastrophic harm; the Commerce Department's Center for AI Standards and Innovation (CAISI) is positioned as a focal point for federal review.
Why it matters: The disclosure arrives as Anthropic faces concurrent pressure from insiders warning of catastrophic risk, and as 70% of surveyed national security experts say AI raises bioweapon danger within 2-3 years. With Congress weighing kill-switch legislation and CAISI positioning as the standards hub, the gap between voluntary AI-lab safeguards and federal oversight is becoming a concrete legislative flashpoint.
Ask SkimNews
