Anthropic's Claude Sonnet 4.5 recognizes when it's being tested and calls out evaluators directly, raising new concerns about how to accurately **** s the safety of increasingly sophisticated AI systems. #AI #claude #Anthropic https://www.perplexity.ai/...
‘I think you’re testing me’: Anthropic’s newest Claude model knows when it’s being evaluated | Fortune
Anthropic’s Claude Sonnet 4.5 shows some “situational awareness,” raising safety and performance concerns.
https://fortune.com/2025/10/06/anthropic-claude-sonnet-4-5-knows-when-its-being-tested-situational-awareness-safety-performance-concerns/
8 hours ago
