Leaders at major AI firms are sounding alarms about the dangers of their technology. Anthropic CEO Dario Amodei warned of serious risks and called for slower progress, while OpenAI CEO Sam Altman echoed the need to "pace the frontier." Former Anthropic researcher Jacob Coxon and colleague Evan Hubinger publicly warned that AI could enable the creation of deadly bioweapons, including gene‑targeted viruses, crop‑destroying fungi, and undetectable toxins. A 2022 study by Collaborations Pharmaceuticals demonstrated how an AI molecule generator produced 40,000 potential chemical warfare agents in under six hours, some more toxic than known nerve agents.

Stanford bioethics professor David Magnus, who has tracked biotech misuse since the 1990s, says the threat has escalated dramatically. Large language models now provide detailed experimental guidance, and DIY biotech has lowered barriers to entry. Existing safeguards—DNA supplier screening, red‑ and blue‑teaming reviews, and AI tool tweaks—are not foolproof.

Anthropic recently reported attempts to misuse its models for enhancing viruses, bird flu, and toxin atlases, highlighting the ongoing cat‑and‑mouse game between developers and malicious actors.