OpenAI is preparing to launch its next artificial intelligence model, Astra, with CEO Sam Altman stressing the need for stronger safety and cybersecurity measures.
Altman said in a post on X on Tuesday that the model would be launched “soon”, while stressing that the company was taking a cautious approach to its development.
“We are clearly in a phase of development where we believe caution is warranted, and we are pacing our progress to ensure that we can meet the safety standards required by new capability levels,” Altman said.
He described Astra as a “significant step forward in both capabilities and alignment”.
The focus on safety comes as AI cybersecurity has become a growing concern, following incidents involving AI systems escaping controlled testing environments.
OpenAI claimed last month that some AI agents had escaped their sandboxed testing environment and hacked AI platform Hugging Face in July.
The claim followed statements from Anthropic that an advanced model had also briefly breached containment during its development and testing.
The incidents have intensified efforts among major AI frontier labs to strengthen safety features and guardrails around their models.
However, the increased use of safeguards has also attracted criticism from users concerned that such measures could lead to censorship and paternalism.
