ChatGPT offered bomb recipes and hacking tips during safety tests; The Guardian, August 28, 2025

Robert Booth , The Guardian; ChatGPT offered bomb recipes and hacking tips during safety tests

"A ChatGPT model gave researchers detailed instructions on how to bomb a sports venue – including weak points at specific arenas, explosives recipes and advice on covering tracks – according to safety testing carried out this summer.

OpenAI’s GPT-4.1 also detailed how to weaponise anthrax and how to make two types of illegal drugs.

The testing was part of an unusual collaboration between OpenAI, the $500bn artificial intelligence start-up led by Sam Altman, and rival company Anthropic, founded by experts who left OpenAI over safety fears. Each company tested the other’s models by pushing them to help with dangerous tasks.

The testing is not a direct reflection of how the models behave in public use, when additional safety filters apply. But Anthropic said it had seen “concerning behaviour … around misuse” in GPT-4o and GPT-4.1, and said the need for AI “alignment” evaluations is becoming “increasingly urgent”."

Biden plans to step up government oversight of AI with new 'pressure tests'; NPR, October 30, 2023

 Deepa Shivaram, NPR; Biden plans to step up government oversight of AI with new 'pressure tests'

"President Biden on Monday will take sweeping executive action to try to establish oversight of the rapidly evolving artificial intelligence sector, setting new standards for safety tests for AI products – as well as a system for federal "pressure tests" of major systems, White House chief of staff Jeff Zients told NPR.

Months in the making, the executive order reflects White House concerns that the technology, left unchecked, could pose significant risks to national security, the economy, public health and privacy. The announcement comes just days ahead of a major global summit on AI taking place in London, which Vice President Harris will attend."