AI models can now design dangerous viruses, raising bioweapon concerns
AI systems already help design viruses and toxins; Anthropic blocked several attempts to weaponise pathogens.
Artificial intelligence can now help people design viruses and pathogens that do not exist in nature. In April, experts warned chatbots give disturbingly easy advice on making pathogens more lethal.
In August, scientists reported they had used AI to build entirely new viruses not seen before. Anthropic said it had blocked several attempts to use its models by people who might have been trying to weaponise pathogens.
The threat from current AI models is still modest, says synthetic biologist Drew Endy from Stanford. But as AI systems specialise in biological research, the risk could grow sharply.
These new models may eventually create actual bioweapons from AI-designed genomes.
- August
- Month AI built novel viruses
- Blocked weaponisation attempts
- Anthropic action
- 6-12 months
- Timeline for AI takeover risk
- Scientists and AI executives
- Open letter signatories
Why it mattersAI could accelerate creation of novel biological weapons faster than human researchers can build defences against them.
AustraliaAustralia relies on international biosecurity agreements and researchers to monitor emerging AI bioweapon risks, but faces the same escalating threat as other nations.
✓ Claims checked against the source and corrected before publish. checked 10 d ago
