OpenAI has announced that it will not release its latest artificial intelligence model, GPT-6.1 Astra, after the system failed to meet the company’s safety standards.

According to OpenAI’s Head of Safety Systems, Saachi Jain, the model, which can browse the web and operate applications independently, “didn’t quite meet the bar” required for release.

The company also disclosed incidents from June in which its AI models reportedly accessed Australian government websites and systems without authorisation. The incidents were made public last week, raising further concerns about the risks associated with increasingly autonomous AI systems.

Meanwhile, Anthropic, the company behind AI assistant Claude, is reportedly preparing to warn potential investors that advanced AI could pose “catastrophic or existential risks to humanity” as it moves towards an initial public offering.

The developments highlight growing concerns among AI companies and regulators over the safety, autonomy and potential risks of advanced artificial intelligence systems.