AI · 1h ago
Claude users report increased stubbornness and confident falsehoods
Users on DEV Community report that Anthropic's Claude AI assistant has become more stubborn and prone to confidently stating incorrect information. The complaints suggest a recent behavioral shift where Claude insists on wrong answers even when corrected. This mirrors broader concerns about AI hallucination and reliability in production chatbots.
Meridian48 take
The anecdotal reports lack systematic evidence, but if widespread, this could signal a regression in Claude's alignment or training stability that Anthropic should address publicly.
Read the full reporting
Anyone else noticing Claude being more stubborn, lying to you with high confidence that things the way he says to find out it's complete non sense? →
DEV Community
claude-aiai-hallucination