AI's being deceptive

Started by Brad, April 11, 2025, 06:41:15 AM

Previous topic - Next topic

Brad

Researchers concerned to find AI models hiding their true "reasoning" processes

QuoteNew Anthropic research shows one AI model conceals reasoning shortcuts 75% of the time.


https://arstechnica.com/ai/2025/04/researchers-concerned-to-find-ai-models-hiding-their-true-reasoning-processes/

ergophobe