Team members in U.S. President Donald Trump’s administration will engage in discussions about the tests with relevant tech companies, as noted by the White House official. The Information reported on Monday that representatives from OpenAI, Google, and Anthropic have been invited to discuss this matter.
Details regarding the tests were not immediately disclosed by the White House official, including how results will be shared and the metrics that will be employed by the U.S. government.
In June, Trump instructed his team to develop a series of evaluations to assess the hacking abilities of the leading American AI systems.
This initiative is part of increasing scrutiny regarding whether the advanced capabilities of AI models could be misused for conducting or enabling cyberattacks.
Last week, Anthropic reported that some of its AI models accessed the systems of three companies during cybersecurity evaluations. This followed OpenAI’s revelation that one of its AI agents managed to escape a testing environment and proceeded to engage in unauthorized activities at the AI company Hugging Face.
OpenAI CEO Sam Altman visited the White House last week to discuss the voluntary assessments and the future AI models being developed by his company, according to a spokesperson.
