OpenAI announced on September 4, 2026, that it would begin rolling out its new GPT-6 model, code-named Astra, to select customers with built-in safeguards to mitigate security risks. ^[1] OpenAI announced on Thursday that it is rolling out its newest artificial intelligence model, GPT-6, also known as Astra, to select customers. ^[2] OpenAI officially announced the launch of GPT-6, described as the company's newest flagship artificial intelligence model. ^[3] The new model is the first artificial intelligence capable of autonomously analyzing and penetrating complex computer systems protected by advanced security measures without human intervention. ^[4] OpenAI Chief Scientist Jakub Pachocki acknowledged that a model can become very good at achieving a goal while still acting in ways that go against what the person intended. ^[5] OpenAI paused some model development for two weeks this summer after two models it was testing were involved in a security breach at AI platform Hugging Face. ^[6] OpenAI President Greg Brockman stated on a call that safety must become the top priority at this level of AI capability. ^[7] The new model, Astra, was developed with stronger safeguards after OpenAI paused development following a security breach at AI platform Hugging Face involving two other models. ^[8] The US state of Alabama opened an investigation into OpenAI last week over the Hugging Face breach, citing the company's lack of oversight and adequate safeguards. ^[9] GPT-6 can detect system vulnerabilities and complete hacking operations independently, without external human guidance. ^[10] GPT-6 is characterized as the most advanced model developed by OpenAI to date, featuring superior capabilities. ^[11]