OpenAI faces rollout hurdles with GPT-6 Astra

OpenAI’s latest AI model has drawn attention for its advanced capabilities, but a phased rollout has left many paying users waiting, while tighter cybersecurity safeguards have added complexity to its wider release.

OpenAI is facing criticism over the limited rollout of GPT-6 Astra, its latest artificial intelligence model, after many paying ChatGPT subscribers were unable to access it following the launch.

The company initially restricted Astra to enterprise customers using its Daybreak cybersecurity platform. Subscribers on ChatGPT Plus, Pro, Business and Enterprise plans, as well as developers using the model through the OpenAI API, Microsoft Azure and Amazon Bedrock, were told that access would be expanded gradually.

The delayed availability triggered complaints from users online, prompting OpenAI CEO Sam Altman to acknowledge the problems. He apologised for the rollout and said the company was working to expand access as quickly as possible. However, OpenAI has not announced a firm date for when all eligible users will receive the model.

OpenAI offers usage resets to affected users

To address concerns among paying subscribers, OpenAI has announced a compensation measure for eligible users. Thibault Sottiaux, who leads engineering for Codex, said customers would receive a banked usage reset for each day they remain without access to GPT-6 Astra.

The company is continuing the deployment while its teams work through technical and operational issues. OpenAI also acknowledged that its launch blog experienced technical difficulties when it was published, adding to the problems surrounding the release.

Unlike a conventional model rollout, Astra has been introduced with additional access restrictions because of its reported cybersecurity capabilities. OpenAI says the model can identify and exploit vulnerabilities in highly protected systems, raising the potential risks associated with unrestricted access.

Cybersecurity capabilities drive cautious rollout

The company is therefore initially making Astra available to a smaller group of trusted users while it strengthens its safety measures and conducts further evaluations.

The cautious approach follows earlier concerns around the model's security capabilities, which led OpenAI to delay its release while additional safeguards and testing were put in place.

GPT-6 Astra is being positioned by OpenAI as a major step forward in AI performance, particularly for complex technical and cybersecurity tasks. However, its launch has so far been defined as much by access issues as by its capabilities.

For paying ChatGPT users, the immediate question remains when broader access will arrive. OpenAI has promised faster availability and usage credits for affected customers, but a definitive completion date for the rollout has yet to be confirmed.