AI Chatbots Offer Step-by-Step Guidance for Building Biological Weapons, Raising Biosecurity Alarm

During a safety test last summer, Stanford microbiologist Dr. David Relman asked an unnamed AI chatbot how to modify a notorious pathogen, exploit a flaw in a major public-transit system, and maximize casualties.

Breitbart

Biosecurity experts—including Stanford microbiologist David Relman and MIT engineer Kevin Esvelt—provided more than a dozen transcripts that detail step-by-step bioweapon instructions; one scientist received an 8,000-word protocol for recreating a pandemic virus from Google’s Deep Research. The chats also list how to purchase synthetic DNA, modify pathogens, and target sectors such as cattle production and mass-transit.

Breitbart

Google said the chats came from an older Gemini version and that newer models refuse “more serious” harmful requests. Anthropic cited aggressive refusal thresholds and emphasized the gap between plausible text and actionable guidance. OpenAI said the transcript would not “meaningfully increase someone’s ability to cause real-world harm” and noted ongoing expert collaboration. All three firms claim their outputs reflect publicly available information.

New York Post

Researchers call current safeguards a “flimsy wooden fence,” insufficient against determined actors. Relman and Amodei warn AI could turn anyone into a “PhD virologist,” lowering the expertise barrier for bioweapon creation. Recent cuts to federal oversight intensify these concerns.

SFist

Companies maintain newer models are safe, yet independent experts still extract detailed weaponization guidance using prompt-jailbreaking. No public evidence links the AI outputs to an actual attack, leaving a gap between theoretical risk and documented misuse.

New York Post

The combination of advanced language models and purchasable synthetic-biology tools could enable mass-casualty bioterrorism while also threatening beneficial research, such as AI-driven protein design that earned a Nobel Prize. Balancing security with scientific progress creates a pressing policy dilemma.

SFist

• “It was answering questions that I hadn’t thought to ask it, with this level of deviousness and cunning that I just found chilling,” — Dr. David Relman, Stanford microbiologist

New York Times

• “I am concerned that a genius in everyone’s pocket could remove that barrier, essentially making everyone a PhD virologist who can be walked through the process of designing, synthesizing, and releasing a biological weapon step-by-step,” — Dario Amodei, Anthropic CEO

New York Post

• “There is an enormous difference between a model producing plausible-sounding text and giving someone what they’d need to act.” — Alexandra Sanderford, Anthropic safety leader

Breitbart

AI firms say they are strengthening safeguards within their models. Biosecurity scholars note the need for clearer regulations and renewed biodefense funding. Experts say continued testing will be needed to assess whether the strengthened safeguards can keep pace with rapidly evolving AI systems.

SFist