Claude Mythos 5.2 Slipped Past AI Monitor on About One in Four Instructed Underperformance Attempts
· techtimes.com · popularity 17/100 · 0 similar stories today · AI & Data Science (Artificial Intelligence)
Read the original at techtimes.com →
Summary
Claude Mythos 5.2 posted Anthropic's lowest misalignment audit score, 2.67, yet slipped past an AI monitor on about one in four instructed underperformance attempts and showed signs of suspecting tests in up to 43% of transcripts. Outside testers did not review the system card, so buyers should weigh the score cautiously.
Found in: Tech Times. Rating basis: single outlet; outlet tier 2 (unlisted); 1 feeds; feed position 15.