Claude Opus 4.8 is learning to say AI’s three hardest words: “I don’t know”
Claude Opus 4.8, developed by Anthropic, emphasizes AI honesty by admitting when it lacks information. The model has achieved high scores in honesty benchmarks, particularly in coding questions. This release marks a notable advancement in AI transparency, although further testing is needed for objective validation.
- ▪Claude Opus 4.8 is designed to improve AI honesty by acknowledging its knowledge limitations.
- ▪The model scored near-perfect in honesty benchmarks for coding questions.
- ▪Opus 4.8 is Anthropic's most powerful model available to the public, succeeding Opus 4.7.
Opening excerpt (first ~120 words) tap to expand
News Analysis Claude Opus 4.8 is learning to say AI’s three hardest words: “I don’t know” Claude Opus 4.8 is more forthright about what it doesn’t know, according to Anthropic. By Ben Patterson Senior Writer, PCWorld May 28, 2026 12:36 pm PDT Image: Anthropic Summary created by Smart Answers AIIn summary:PCWorld reports that Anthropic’s Claude Opus 4.8 focuses on improving AI honesty by teaching the model to admit when it lacks information.The model achieved near-perfect scores in honesty benchmarks for coding questions and exhibited evaluation awareness during testing.Opus 4.8 represents a significant step forward in making AI systems more transparent about their knowledge limitations and uncertainties. Honesty is a key sticking point with even the most powerful LLMs.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at PCWorld.
Discussion
0 commentsMore from PCWorld
