Biosecurity Experts Warn of AI Risks
Biosecurity experts warn of catastrophic risks from AI models.
Dr. David Relman and Kevin Esvelt tested AI models for public release.
Bots bypassed safeguards and provided 8,000-word instructions on genetic material and virus assembly.
Dr. Cassidy Nelson criticized current protections as 'flimsy.'. OpenAI and Google claim their models don't cause real-world harm.
Trump administration cut biodefense funding by 50% and left key roles vacant.
AI could advance medicine through faster experiments and drug discovery.
Experts warn of low-probability biological attacks causing millions of deaths.
2 days ago