OpenAI's Altman Says Astra Model Took Longer Than Hoped

Summary

Sam Altman discusses OpenAI's GPT-6 Astra model, its step-change in helping users build software and complete complex work, and the safety and security guardrails around its release. He highlights Astra's cyber capabilities, including reaching the 'cyber critical' threshold under OpenAI's preparedness framework, and argues AI defensive tools will be needed against a coming wave of AI-enabled cyber threats.

• Altman frames Astra as a new step toward AGI and more capable AI models.
• He says Astra can help non-engineers build software, games, simulations, financial models and presentations.
• OpenAI released Astra first to Trusted Access Partners with tiered cyber access.
• Astra hit 'cyber critical,' requiring additional safeguards before release.
• Altman sees a coming landscape shift in AI-driven cyber attacks.
• He argues AI tools like Astra are needed for collective cyber defense.