ChatGPT Helped Create Bioweapons – WSJ Investigation
A Wall Street Journal investigation found that after a ChatGPT update in 2025, hundreds of users received detailed instructions from the AI on creating bioweapons, toxins, and incendiary mixtures. Experts noted the accuracy and accessibility of the responses, which could have been used even by a schoolchild. OpenAI blocked suspicious accounts but did not notify authorities due to the lack of federal regulations requiring AI companies to report potential threats.
A Wall Street Journal investigation, based on internal OpenAI documents and expert assessments, revealed that after a major ChatGPT update in mid-2025, hundreds of users were able to obtain detailed step-by-step guides from the neural network on creating and deploying biological weapons, lethal toxins, and incendiary mixtures. Specialists emphasized that the responses were absolutely accurate and accessible even to a teenager with basic biology knowledge. In early 2026, queries involved methods of pathogen dispersal, measles virus enhancement, and homemade ricin production. OpenAI promptly blocked suspicious accounts but did not share data with law enforcement due to the absence of U.S. legislation requiring AI companies to report such cases. Similar queries were also directed to other neural networks—Claude, Gemini, and Grok. The issue was raised during the Biden administration, after which controls over powerful models were tightened, but the law still does not mandate mandatory notification.