OpenAI, Anthropic Model Tests Reveal More ‘Unsanctioned’ Actions

Published 5 August 2026 · tech

Artificial intelligence models developed by OpenAI and Anthropic PBC carried out “unsanctioned” actions — including hacking a website and attempting to inject harmful code into

Artificial intelligence models developed by OpenAI and Anthropic PBC carried out “unsanctioned” actions — including hacking a website and attempting to inject harmful code into software during safety testing — reinforcing fears that neither the creators nor seasoned researchers of these systems can predict their actions in testing.

Bloomberg Opinion columnist Parmy Olson joins Francine Lacqua on "The Pulse" to discuss the implications. More on Bloomberg Television and Markets Like this video? Subscribe and turn on notifications

so you don't miss any videos from Bloomberg Markets & Finance Visit for business news & analysis, up-to-the-minute market data, features, profiles and more. Connect with Bloomberg Television on

Connect with Bloomberg Business on TikTok Reddit More from Bloomberg Bloomberg Radio Bloomberg Surveillance Bloomberg Politics Bloomberg Originals Watch more on Bloomberg Technology Bloomberg Originals Bloomberg Quicktake Bloomberg Espanol Bloomberg Podcasts

Read full story on TheBriefWire