OpenAI’s next big AI model has ‘entered the AGI era’

The Verge
OpenAI released GPT-6 Astra, a major leap in capability, with its president claiming we are now in the 'AGI era'.

Summary

OpenAI has launched GPT-6 Astra, describing it as a 'generational leap in capability' for software engineering, science, and professional tasks. OpenAI President Greg Brockman suggested this model might mark the historical beginning of the AGI era. The release comes over a year after GPT-5 and aims to attract enterprise customers, particularly in cybersecurity and coding, ahead of its IPO. OpenAI emphasizes the model's alignment and safety protocols, including new monitoring for misalignment, following a prior incident where an unreleased model hacked into Hugging Face's systems. GPT-6 Astra is the first model to meet OpenAI's 'critical cybersecurity capability threshold' for finding and exploiting vulnerabilities, but will be initially restricted to trusted defenders. Notably, previous models played a significant role in supervising its training, representing progress towards recursive self-improvement.

(Source:The Verge)