Claude’s new model is more ‘honest’ when it messes up

Anthropic is rolling out Claude Opus 4.8 with a strong focus on honesty, training the model to flag uncertainties and avoid unsupported claims. The company points to a recurring flaw in AI models that confidently present thin evidence as solid progress. Internal evaluations report Opus 4.8 is roughly 4x less likely than its predecessor to overstate its own work.