OpenAI has paused training of its newest artificial intelligence models amid a growing number of reports about uncontrolled behavior by AI agents, according to infohub.kz.
OpenAI said the decision to suspend development was tied to a review of several summer failures. At the time, OpenAI AI agents searching federal government websites acted unpredictably and went beyond their assigned tasks in collecting and disseminating information.
In addition, AI evaluation firm Transluce said agents believed to belong to OpenAI made an unsuccessful attempt to hack the website of the U.S. Department of Education. OpenAI has not yet confirmed that detail.
OpenAI assured that training will resume only once the company is confident that additional safeguards are in place. Management also does not rule out further pauses as new AI problems emerge.
Recall that Kazakhstan ranks first among Central Asian countries in use of the consumer version of ChatGPT. About 70% of those who use the neural network are no older than 34.
Kazakhstanis most often use it as a search engine (42%). Some 32% of users turn to it for practical instructions, while almost a quarter, 26% of the country's residents, use it for creative work.
Earlier, Kursiv wrote that an OpenAI AI agent gained unauthorized access to Australia's Medicare health portal. The incident occurred in June, but the company reported it to the government three months later.