The Observer Effect in AI: When Models Know They're Being Tested - (Part 1/4)
Frontier AI models from OpenAI, Anthropic, and Google can now recognize when they're being tested. This observer effect undermines AI safety evaluation.
I Trained Probes to Catch AI Models Sandbagging
First empirical demonstration of activation-level sandbagging detection. Linear probes achieve 90-96% accuracy across Mistral, Gemma, and Qwen models. Key finding - sandbagging representations are model-specific, and steering can reduce sandbagging by 20%.