OpenAI co-founder Greg Brockman stated in an interview with a16z that humanity has entered the era of Artificial General Intelligence. He noted that models like GPT-6 Astra, which possess long-term autonomous execution and computer operation capabilities, mark the crossing of this threshold.
Brockman emphasized that the focus is shifting from predicting when AGI will arrive to managing the deployment of increasingly powerful systems. He defined AGI as a highly autonomous system capable of outperforming humans in most economically valuable tasks.
To address security, OpenAI reallocated 25% of its production engineers to focus on system defense and architecture upgrades. The team utilized the Astra model to identify vulnerabilities within internal systems, successfully patching multiple critical issues.
The company reported that vulnerability detection reached a saturation point, with the team identifying all P0-level and critical issues that Astra could recognize. Brockman noted that this does not imply the absence of other flaws in the system.
Looking ahead, the team plans to initiate new rounds of security audits as future models are developed. OpenAI intends to deploy next-generation cybersecurity capabilities internally to detect previously unidentified vulnerabilities.
