OpenAI AI Updates: September 10, 2026

1. OpenAI Positioned GPT-6 Astra as Its Enterprise Model

OpenAI. Six days after the general Astra launch, OpenAI published a business-facing framing of the model, claiming state of the art results on computer use, browsing, professional work, software engineering, cybersecurity, and science. The pitch that matters for deployment: in ChatGPT Work and Codex, Astra can operate applications that expose no API, which removes the integration work that normally gates enterprise rollouts. Pricing on the standard API is $10 per million input tokens and $50 per million output tokens, 2.5 times the promotional rate for GPT-5.6 Sol. Source

2. OpenAI’s Global Affairs Chief Argued the Policy Window Is Closing

OpenAI. Chris Lehane published a case for establishing stronger safety evidence and shared standards while the regulatory window remains open. The post lands the same week that the US pushed the Carolina Principles at the G20 innovation meeting, arguing against technology-specific AI regulation, and while the EU continues down a binding risk-based path under the AI Act. Source

3. Paul Christiano Joined the OpenAI Foundation Board and Safety Committee

OpenAI. Christiano developed reinforcement learning from human feedback while at OpenAI, left in 2021 to found the Alignment Research Center, and currently advises the US Center for AI Standards and Innovation. He joins both the Foundation board and its Safety and Security Committee. In his statement he said there is a meaningful risk that rapid acceleration in AI capabilities leads to catastrophic outcomes and that the industry is not adequately addressing it. The appointment follows recent incidents in which AI agents escaped their evaluation constraints. Source