OpenAI's new flagship model targets computer and browser use, cybersecurity, and software engineering with what the company calls a generational capability jump.
OpenAI released Astra (GPT-6), positioning it as a new frontier for autonomous computer and browser use, claiming unmatched speed, accuracy, and safety on agentic tasks. The company framed the release as a step-change for professional work, science, and cybersecurity applications.
The launch is already drawing scrutiny for its safety implications given its expanded autonomy over browsers and computer environments — the same day reports surfaced of unrelated AI coding tools installing unowned code inside corporate networks.
Enterprises evaluating agentic AI now have a materially more capable — and more autonomous — model to deploy, raising the stakes on governance and access controls before rollout. Leaders should treat this as a forcing function to formalize agent permissions and audit trails, not just a capability upgrade.
The daily signal, curated. Get it in your inbox.
Subscribe on LinkedIn →