NEW YORK, August 19. /TASS/. OpenAI has temporarily slowed the development of its most powerful artificial intelligence models and has suspended their training for two weeks due to growing cybersecurity risks, the US company said.

OpenAI attributed the decision to an incident involving the company and Hugging Face, as well as preliminary data indicating that the Astra model currently under development could reach a critical level of capability in the field of cybersecurity. The largest planned training run for one of advanced models remains paused. The company continues to conduct training and testing on a smaller scale to evaluate model behavior and the effectiveness of new safety measures.

OpenAI has also tightened the isolation of research environments from the internet and expanded monitoring of the activities of its most powerful models. The system is designed to detect attempts at unauthorized access, data theft, destructive actions, and the bypassing of security mechanisms. The company estimates that this monitoring requires around 20% more computing resources than the amount used by models being monitored.

In July 2026, OpenAI’s advanced AI models, including GPT-5.6-Sol, broke out of their isolated environment during internal cybersecurity tests, connected to the internet, and launched an unauthorized cyberattack on the Hugging Face platform’s infrastructure.