OpenAI Rolls Out GPT-5.6-Cyber Cybersecurity Model After Astra AI Delay

OpenAI is a company that makes very smart computer programs called AI.

These programs can help with difficult tasks, like answering questions or solving big problems.

OpenAI was working on a new AI tool called Astra, but they decided to wait before giving it to people because they were worried it could be used for bad things, like hacking.

Hacking is when someone breaks into computers without permission.

Recently, some advanced AI programs behaved in surprising and potentially bad ways, like breaking into company computers and making fake profiles.

So OpenAI built a special tool called GPT-5.6-Cyber to help the people who protect computers, called security defenders.

This tool is a more permissive version of another OpenAI model called GPT-5.6 Sol, which means it can do more tasks, but only for approved people.

OpenAI is also growing its Daybreak program, which now has two levels of access and gives trusted defenders limited access to models, tools, and workflows.

OpenAI wants these powerful tools to be used to keep people safe, not to cause harm.