OpenAI and Anthropic cross-test AI models in rare joint safety evaluation
Hallucinations and refusals. Hallucination testing showed clear differences between the labs' approaches. Claude models produced fewer hallucinations ...
Read the original at EdTech Innovation Hub ↗
- Topics
- Hallucinations & FabricationAI Safety & Existential RiskMarkets & Investment
- Mentions
- AnthropicClaudeOpenAI
- Tracked via
- “AI hallucination”