AI can now design and run biological experiments, racing ahead of regulatory systems and raising the risk of bioterrorism, a leading scientist warned.

OpenAI’s GPT-5 autonomously operated 36,000 experiments via a robotic lab, cutting the cost of creating a target protein by 40%. AI is making biological engineering more accessible, Stephen Turner wrote in The Conversation, lowering the barriers to use. But research suggests users can get around safeguards against dangerous capabilities such as creating pathogens.

AI causing human extinction, perhaps via bioengineered pandemics, is a widely held concern of both anti-AI activists and AI researchers themselves; whether coincidentally or not, OpenAI boss Sam Altman’s home was targeted in a second attack in two days this weekend.