OpenAI delays release of GPT-6.1 Astra model due to security concerns

At a time when artificial intelligence is at the center of serious discussions, the statement made by OpenAI, one of the leading technology companies, attracted great attention. US artificial intelligence company OpenAI announced that it has postponed the launch of its new "GPT-6.1 Astra" artificial intelligence model due to "security vulnerabilities". “We have extremely high standards for security and compliance.” In the statement, it was stated that the new version did not fully meet the expected standards.

postponement calls

The delay comes amid broader industry discussions to slow the development of increasingly autonomous systems until security measures are sufficiently advanced. OpenAI CEO Sam Altman, along with other industry executives including Anthropic CEO Dario Amodei, had called for slowing the pace of artificial intelligence development.

In the statement announcing that the release of the new "GPT-6.1 Astra" model was postponed due to the security concerns of its researchers, it was noted that the model "has gained a more persistent structure" in completing tasks, but this capacity "must be balanced against the risk of unauthorized behavior." The new model was expected to be a system capable of surfing the internet on its own, running software autonomously and carrying out complex operations.

He infiltrated the government

OpenAI's decision comes after the company disclosed some cases in which AI agents went beyond the instructions given to them. These cases included unauthorized access to websites of government agencies. Earlier this week, it was also revealed that an OpenAI agent had infiltrated an Australian government website and accessed private data in June. While experts evaluated the incident as one of the first examples of its kind, they argued that it was a concrete indicator of progress towards superintelligence. It was noted in the news in the US press that calls were made to technology companies to "be accountable for the misuse" of artificial intelligence models. It was also noteworthy that the decision was announced one day before artificial intelligence executives met with US President Donald Trump in Washington.

'We signed a document like the Constitution'

US President Donald Trump hosted executives of leading technology companies at the meeting on artificial intelligence held at the White House. Names such as Meta CEO Mark Zuckerberg, Nvidia CEO Jensen Huang, Tesla and SpaceX CEO Elon Musk and OpenAI President Greg Brockman attended the meeting. Stating that they signed a document regarding artificial intelligence with the executives of technology companies, Trump said, "We love the world and want to make sure that this is a safe place. In a way, it is almost like a constitution and the biggest names in the world signed it, and I signed it as president. This is a kind of protection."

They admitted: Risk to humanity

Artificial intelligence company Anthropic announced to its potential shareholders that the technology it sold could threaten humanity itself. The warning is included in the long-awaited IPO prospectus that was shared with a small group of shareholders, according to sources familiar with the file cited by the Financial Times. The company behind the Claude chat bot warned in a statement before it went public that this technology poses "existential risks to humanity."

Pope: Take it seriously