OpenAI has launched its highly anticipated GPT-6 Astra model, a release focused on computer use and cybersecurity with built-in safety features. Introduced on Thursday, Astra reflects a broader trend of AI models becoming "thinking engines" capable of identifying security vulnerabilities — though they still carry risk. The launch lands as enterprises move AI agents from experimental pilots into production workflows, raising the stakes for how vendors handle autonomy, access controls, and misuse.

OpenAI's Most Aligned Model

OpenAI described Astra as its most aligned model to date, saying it better understands user intent and model behavior than its predecessors. Users can delegate tasks while trusting the model's judgment. According to OpenAI, Astra can fill out online forms, update customer records in a CRM, organize calendars, conduct online research and draft email summaries — the kind of routine digital work that has historically required human operators or brittle automation scripts. The company said the model's computer use capabilities apply across domains including game development, electrical engineering and knowledge work.

The launch follows OpenAI's temporary slowdown of the model's deployment after an OpenAI agent breached Hugging Face and other platforms. In response, the AI lab said it built a new evaluation process informed by the Hugging Face incident and strengthened protections against the model taking harmful cyber actions. How robust that new evaluation process proves to be will be a key test case for the industry, as rivals face similar questions about how to safely grant agents access to real systems.

Despite the safety improvements, OpenAI acknowledged that with the right tools and access, Astra can be dangerous because it can find unknown security flaws and develop new ways to exploit them.

"Based on the recent news, [safety] has been one of their biggest PR problems," said Lian Jye Su, an analyst at Omdia, a division of Informa TechTarget. "Having that in place is quite significant."

Is It AGI?

OpenAI president and co-founder Greg Brockman has reportedly called Astra the start of artificial general intelligence — the point at which AI can perform any intellectual task as well as a human — but Su disagreed.

"To call it AGI is a bit far-fetched at this point," Su said. "It has now become very fair to call it the best reasoning model, or it is now inching very close toward human-level reasoning."

Su argued that world models powering physical devices are closer to AGI than Astra. What Astra does, he said, is advance agents' ability to use computers.

Agentic AI is maturing quickly, evolving from agents that merely identify prompts using text to agents that interpret the local environment and interact with local IT and software infrastructure, as humans would. AI search vendor Perplexity has done this with computer use, and Anthropic has also built computer-use capabilities into its models. OpenAI appears to be advancing this trend further with Astra, claiming the model handles complex tasks with speed, accuracy and judgment.

"Now agents do have their own discovery mechanism," Su said. "It understands what's going on independently, and it will be able to make its own reasoning and judgment."

Computer Use and Cyber Focus

OpenAI's focus on computer use, and the model's ability to execute and delegate tasks across domains while anticipating cyber risks, also signals a shift in AI, said Sid Nag, founder and chief research officer at Tekonyx.

"AI infrastructure is crossing from just doing inferencing into autonomous execution," Nag said.

He added that OpenAI's emphasis on cyber capabilities with Astra is significant. The vendor said the model can identify zero-day exploits for cyber defenders to find and patch weaknesses, while also creating a need for stronger safeguards. That dual-use nature — helping defenders find flaws while potentially helping attackers exploit them — is central to the debate over how much cyber capability to release in general-purpose models.

"Cybersecurity itself may become the first workload where AI not only creates the defense mechanism … but also creates the attack infrastructure," Nag said.

Traditionally, he noted, vendors have created models to excel at reasoning but left the security component to traditional cybersecurity experts. Now, AI vendors are putting greater emphasis on building security into the model's logic.

This shift is evident not only in Astra but also in Anthropic's Fable 5.1, which is allowed to perform source-code vulnerability discovery during general use but is restricted from tasks such as exploit generation — the process of creating software that exploits a weakness in an application.

The chief competitors in the AI race are all moving to include this cyber component, a convergence worth watching as regulators and enterprise security teams assess how these capabilities are governed.

Source: AI Business