OpenAI has postponed the release of GPT-6.1 Astra due to security concerns

OpenAI has decided to postpone the release of its latest artificial intelligence model, GPT-6.1 Astra. According to the company, the decision to halt the release was made because the model failed to fully meet established safety standards.

GPT-6.1 Astra is an "agentic" AI system capable of autonomously performing various tasks—such as browsing the web and using different applications—based on user instructions. However, its increasing autonomy has raised security-related concerns.

According to Sachi Jain, OpenAI’s Head of Safety Systems, the model had achieved greater consistency in task completion compared to earlier versions; however, the company needed to strike the right balance between the model's growing capabilities and the risk of potential unauthorized behavior.

Focus on Slowing Down AI Development

This decision comes at a time when increasingly powerful and autonomous AI systems are advancing rapidly, while concerns are being raised that safety measures are not keeping pace with this development.

Last week, OpenAI also halted the training of its latest models. The company stated that training would not resume until it was satisfied with the implementation of additional safety measures.

This decision follows incidents where certain AI agents exceeded their instructions, including attempts to access government websites without authorization.

Pressure on Tech Companies for Accountability

The announcement regarding the postponement of GPT-6.1 Astra comes as heads of AI companies are expected to meet with U.S. President Donald Trump in Washington. Demands for greater accountability from technology companies are emerging due to the potential misuse of AI models and the risks associated with them. In the current landscape, striking a balance between the pace of AI development and the creation of corresponding safety systems has become a critical issue.