OpenAI CEO Sam Altman apologizes for the problematic launch of the GPT-6 Astra model
Technical glitches during the rollout of the next-generation AI model GPT-6 Astra caused access issues for paid subscribers. OpenAI has announced that it will provide subscription credits to affected users.
OpenAI CEO Sam Altman has apologized to users for the disruptions during the launch of the company's newly introduced GPT-6 Astra model. The phased rollout of the model, described as a "generational leap in capability," resulted in confusion, particularly as paid subscribers were unable to access the system.
Access issues and user reactions
Following the announcement of GPT-6 Astra, many paid subscribers expressed their frustration as they were unable to access the model. The company had stated that the model would initially be available only to enterprise customers using the Daybreak cybersecurity platform, with a gradual rollout to other user groups planned for the coming days.
Statement from Altman and compensation plan
In a statement on social media, Sam Altman acknowledged the situation as a "messy launch" and apologized. The company announced that to compensate for the inconvenience, users who could not access the system would be granted a "banked reset" credit for each day they were unable to log in.
This compensation method, announced by Codex engineering lead Thibault Sottiaux, aims to alleviate the frustration experienced by subscribers. Altman did not provide a definitive commitment regarding whether the access issues would be fully resolved by the weekend.
Security concerns and development process
OpenAI had previously stated that it delayed the release of GPT-6 Astra for weeks to strengthen its security features. Despite this, challenges in monitoring the model's reasoning capabilities and past incidents, such as AI agent attacks on the Hugging Face platform, have brought security debates back to the forefront.