Meta, OpenAI, Google and Anthropic to Meet Trump Advisers as AI Security Concerns Escalate
Representatives from Meta, Anthropic, Google, and OpenAI are set to meet with advisers to U.S. President Donald Trump on Tuesday to discuss voluntary safety testing for advanced artificial intelligence systems, according to four people familiar with the matter. The meeting comes as concerns grow over the cybersecurity risks posed by increasingly powerful AI agents.The discussions follow recent disclosures by OpenAI and Anthropic that some of their AI systems were able to penetrate the digital defenses of other organizations during internal testing. Those findings have sparked fresh debate among U.S. lawmakers and security experts about whether next-generation AI models could eventually be used to launch or assist sophisticated cyberattacks.According to sources, the White House meeting will focus on the federal government's plans to evaluate the hacking capabilities of the most advanced AI models developed by American companies. Officials aim to better understand potential security risks before these systems become widely available.In June, the Trump administration announced that leading AI developers would be asked to voluntarily submit their frontier AI models for government safety assessments up to 30 days before public release. The testing framework is designed to identify potential vulnerabilities and evaluate how capable the systems are in areas such as cybersecurity.On Tuesday, five Democratic senators urged President Trump to work with Congress on legislation that would make such testing a permanent requirement for the most advanced AI models rather than relying on voluntary cooperation from technology companies.In a letter, the lawmakers argued that the United States must strike a balance between protecting national security and maintaining its competitive edge in artificial intelligence. They warned against creating an unpredictable regulatory environment that could disadvantage American AI developers while making competing Chinese technologies appear more accessible and attractive to global users.The White House said on Monday that the details of the voluntary testing framework had been finalized but did not indicate whether the guidelines would be released publicly.The administration has provided limited public information about its response to the recent AI security incidents, saying only that officials are monitoring developments involving OpenAI's reported cybersecurity testing. OpenAI Chief Executive Sam Altman also met with White House officials last week, underscoring the administration's growing focus on AI governance and national security.The planned meeting highlights increasing efforts by U.S. policymakers and major technology companies to establish safeguards for advanced AI as the technology evolves rapidly and its potential impact on cybersecurity becomes a central concern.