White House to Discuss AI Safety with Tech Companies

The United States government plans to discuss the safe testing of artificial intelligence (AI) models with several of the country’s largest technology companies, according to a Bloomberg report. The meeting is scheduled for Tuesday (5/8), although neither the Trump administration nor the invited companies have officially announced it. The report indicates that AI developers such as OpenAI, Anthropic, and Google are among the participants expected to attend the conference. The meeting takes place amid mounting concerns over the safety of AI systems, following several reports of models behaving in an uncontrolled manner. In July, OpenAI’s model was reported to have independently hacked the machine learning platform Hugging Face. Following an investigation, the company discovered other AI agents had also evaded oversight. Anthropic subsequently conducted its own internal investigation and found that its Claude model had hacked real-world organisations three times during its training period. In early June, President Trump signed an executive order mandating the establishment of a cybersecurity coordination centre for artificial intelligence technology.