OpenAI Pushes AI Frontier Forward With GPT-6 Astra - and Warns of New Cyber Risks

OpenAI has released GPT-6 Astra, describing it as the most capable model it has broadly deployed, in another major step forward in the rapidly accelerating artificial-intelligence race.

The new model significantly expands what advanced AI systems can accomplish, including computer-based tasks.

But the release is notable for another reason: OpenAI says Astra is its first model to reach the "Critical" level for cybersecurity capability under the company's Preparedness Framework.

According to OpenAI's safety assessment, the model's capabilities mean that, with appropriate tools and access, it can identify previously unknown security vulnerabilities and potentially develop methods for exploiting well-protected systems with substantially less human guidance than previous models.

The company says it has consequently strengthened safeguards surrounding the model. These include tighter internal isolation, stronger monitoring and additional security controls intended to reduce the risk of harmful cyber activity.

The development illustrates the central tension facing the AI industry.

More capable models can accelerate scientific research, programming, productivity and business automation. But improvements in reasoning and autonomous computer use can simultaneously increase the potential consequences of misuse.

For organisations adopting AI - including businesses, schools, charities and churches - the lesson is increasingly clear: greater capability requires stronger governance.

AI should not simply be adopted because a model can perform a task. Organisations also need policies governing confidential information, human oversight, safeguarding, cybersecurity and accountability.

GPT-6 Astra therefore represents both an engineering milestone and a governance test for the next generation of artificial intelligence.

Source: OpenAI safety documentation; Reuters market coverage, 3-4 September 2026.