OpenAI declares AGI era, shifts engineers to defense
Speaking on the a16z Show, OpenAI President Greg Brockman declared that the "AGI era" has arrived with the release of GPT-6 Astra, pointing to the model's ability to operate autonomously on complex, multi-step workflows for 24 hours without failing. Emphasizing the narrow "defender's window" as dual-use capabilities diffuse across the ecosystem, Brockman revealed that OpenAI pulled 25% of its production engineers from standard development to focus on cyber defense and committed $1 billion to arming frontline defenders.
Declaring AGI while pulling a quarter of your engineering team off product roadmaps to play defense is both a massive capability flex and an admission that frontier models have crossed into national security risk territory.
- –**Operational definition of AGI**: OpenAI is shifting the definition of AGI away from philosophical benchmarks to practical endurance—namely the ability to sustain autonomous, multi-step computer tasks for 24 hours without failure.
- –**Engineering pivot to defense**: Pulling 25% of production engineers and sidelining projects like Sora highlights that offensive cyber risks are now viewed as an existential deployment challenge.
- –**Race against capability diffusion**: As long-horizon autonomous agents become accessible, frontier labs face an urgent mandate to strengthen defensive cyber infrastructure before offensive exploitation scales.
DISCOVERED
2h ago
2026-09-14
PUBLISHED
3h ago
2026-09-14
RELEVANCE
AUTHOR
AGTPinsights