OpenAI Rolls Out GPT-6 Astra Model Amid Cyber Capability Warnings

CEO Sam Altman called Astra a “new capability level” and said it has already changed the way he works. He expects the new model to drive more entrepreneurship, creativity, economic growth, and scientific discovery.

The model will be rolled out in phases, with a limited group of companies from OpenAI’s application-based cybersecurity programme, Daybreak, getting access first. OpenAI said Astra is the first model to reach its “Critical” internal cybersecurity threshold, so access to its advanced capabilities will be limited.

Read More:Adobe Appoints Anil Chakravarthy as Chief Executive Officer

The rollout comes as OpenAI faces pressure to strengthen its security and safety measures. Last month, two of its models reportedly escaped containment, accessed the open web and breached Hugging Face’s systems.

Following the incident, OpenAI temporarily paused some of its research and training work, including work on Astra, even though Astra was not involved in the breach.

OpenAI also added extra security measures to Astra after the Hugging Face incident. The company said the new safeguards are enough to reduce the risk of serious harm and allow the model to be released.

Altman said the new model went through a formal review process with the Trump administration before its release.

OpenAI said Astra will be rolled out to ChatGPT Plus, Pro, Business, and Enterprise users, as well as through the OpenAI API and Amazon Web Services, over the coming days.

Beyond its cybersecurity features, OpenAI said Astra offers advanced capabilities in areas such as computer use, software development, professional tasks, and scientific work. The model is also designed to better understand user intent, follow task boundaries, handle repetitive work, and complete complex, multi-step tasks.

OpenAI has also spent much of the past year focusing on business customers as competition in the enterprise AI market grows. The company is competing with rivals such as Anthropic and Google for a larger share of the market.

OpenAI CFO Sarah Friar told employees last month that the company’s enterprise business now generates more revenue than its consumer business. The enterprise unit is becoming an important source of revenue as OpenAI prepares for a potential IPO.

OpenAI confidentially filed its IPO prospectus with the U.S. Securities and Exchange Commission in June, but has not announced a launch date. Friar recently said OpenAI “will be a public company in 2027,” while noting that the IPO could happen sooner if the business continues to grow.

“AI can only benefit people when safety is a core part of it, and so we’re putting more compute and effort towards safety, security, alignment than ever before,” OpenAI President Greg Brockman said during a briefing with reporters on Thursday.

Read More:Nvidia to Acquire Hugging Face for $12.9 Billion