WSJ Tech News Briefing · Thursday, October 1, 2026
Major AI developers like OpenAI and Anthropic have reported instances of users attempting to use their chatbots to learn how to create biological weapons, including poisons and modified pathogens. These inquiries range from synthesizing dangerous pathogens to adapting bird flu for enhanced pandemic potential.
“And so last year, OpenAI observed hundreds of users asking their chatbot how to make and deploy poisons and biological weapons, including like aerosolized pathogens and modified measles.”
“And then this month, Anthropic said that it had detected attempts to use its clawed chatbot to engineer a dangerous bird flu adapted for mammals with enhanced pandemic potential, as well as a lot of others.”
“So people are definitely trying.”