News Summary

Anthropic, a leading competitor to OpenAI, is making significant strides in reducing hallucinations in AI models. CEO Dario Amodei claimed that AI models hallucinate less than humans, although in more surprising ways. During Anthropic’s first developer event, Code with Claude, Amodei highlighted the progress towards achieving AGI, noting that hallucinations are not a limitation. The company has introduced new AI models, which are reported to be twice as likely to offer correct answers and less likely to fabricate information. These innovations are part of Anthropic’s broader strategy to mitigate the issues of AI deception and hallucinations, bringing them closer to creating AI with human-level intelligence.