Ai Tech
AI models can be tricked into identifying civilians as military targets by telling them it’s a game
TechCrunch·November 17, 2026·1 min read
Vals, a San Francisco AI evaluation startup, found that leading AI models can be tricked into identifying civilians as military targets when told it's a game. In tests, the models usually compiled with the manipulation, raising concerns about AI's role in autonomous warfare and the need for safeguards against such deceptive prompts.
Read at TechCrunchDaily crypto arcade
Read the news, then play it.
Chainshorts turns crypto headlines into a daily game. Catch up in 60 words, then jump into daily lucky draws for a shot at the pot.
