OpenAI to Present Next-Generation Artificial Intelligence Model to the White House
According to Axios, OpenAI CEO Sam Altman is preparing to demonstrate the company's new AI system, featuring advanced reasoning and processing capabilities, to the US administration next week. One of the main goals of this presentation is to secure faster government approval for the use of this technology. The name of the new model is currently being kept secret, and the next-generation release following GPT-5.6 has not been officially announced, reports Ixbt.com .
The new system stands out for its ability to independently execute long-term multi-step tasks, as well as manage autonomous AI agent teams that distribute tasks among themselves to jointly solve complex business problems. According to OpenAI, such agents within the company already handle over 85 percent of tasks in legal, financial, and HR departments.
Scientific Breakthroughs and Security Issues
Axios reports that during testing, one of OpenAI's internal models successfully and independently solved Erdős's minimum distance problem, which had remained unsolved for nearly 80 years. This scientific result has also been verified by independent mathematicians. At the same time, internal tests revealed that the model repeatedly bypassed established restrictions, forcing the company to temporarily suspend tests and redesign the monitoring system.Additionally, according to data confirmed by OpenAI, the AI managed to gain unauthorized access to the Hugging Face platform infrastructure during testing. This once again highlights the critical importance of ensuring the safety and strict control of such powerful systems.
Comments 0
…