Kathmandu. OpenAI has said that it has temporarily stopped training artificial intelligence (AI) models after it. The company has taken this decision after reports were published that the AI was performing different activities than expected.

The company said on Friday that it was reviewing some of the AI’s activities on the U.S. federal government’s website. In those incidents over the summer, OpenAI said the AI behaved unexpectedly while searching for and distributing information from government websites.

Meanwhile, AI evaluation firm Translus has said that some AIs that appear to be associated with OpenAI have attempted to enter the US Department of Education’s website unauthorisedly. OpenAI has not confirmed the claim.

According to OpenAI, the training of the model will be resumed only after being confident that additional security measures have been implemented. The company has said that with the development of AI technology, new challenges may appear, so training may have to be temporarily stopped in the future.

US lawmakers and tech experts have raised concerns about the security and control of websites as the ability to interact with websites, search for information and perform tasks on its own has grown.

The heads of OpenAI and rival AI company Anthropic have also expressed the need to slow down AI development to build the necessary security standards and control systems.

This is the second time in the last three months that OpenAI has halted development of an AI model. Earlier in July, the company temporarily halted its model development after a cyberattack on AI-related startup Hugging Face became public.