Test shows AI models likely to give advice for terrorist attacks
First reported 9 Oct, 03:00 UTCUpdated 17h ago3 sources
1 left · 1 center · 1 right · Full breakdown
All but two of 134 leading large language artificial intelligence models gave information useful to launching a mass casualty attack or building deadly weapons, in a test of the industry's safety measures. A stress test of the industry found that safety measures can be stripped from leading models within three days…
Excerpt from The National (UAE)
Compare the headlines
How outlets on each side framed it
UK nonprofit Tech Against Terrorism tested over 130 AI models with hundreds of simulated terrorist attack
All but two of 134 leading large language artificial intelligence models gave information useful to launching a mass casualty attack or building deadly weapons, in a test of the industry's safety measures. A stress test of the industry found that safety measures can be stripped from leading models within three days…
Research finds models stripped of safety safeguards consistently provide responses to potentially dangerous requests, media reports
Full coverage
3 outlets — tap a headline to read it at the source
- Test shows AI models likely to give advice for terrorist attacks (opens The National (UAE) in a new tab)
All but two of 134 leading large language artificial intelligence models gave information useful to launching a mass casualty attack or building deadly weapons, in a test of the industry's safety measures. A stress test of the industry found that safety measures can be stripped from leading models within three days…
- 3 in 5 AI models fail terrorism safety tests: Study (opens The Express Tribune in a new tab)
UK nonprofit Tech Against Terrorism tested over 130 AI models with hundreds of simulated terrorist attack
- 3 in 5 AI models fail terrorism safety tests: Study (opens Anadolu Agency in a new tab)
Research finds models stripped of safety safeguards consistently provide responses to potentially dangerous requests, media reports
