The Office of the National Cyber Director has distributed a set of questions to various tech firms regarding how to protect against AI-powered cyber threats, POLITICO reported, citing four individuals familiar with discussions between the administration and the technology sector. Companies have been instructed to submit their responses by Friday.
Some of these questions were explored during a Tuesday afternoon gathering at the White House involving cyber officials and approximately 30 industry representatives, one source said. The participants, who spoke on condition of anonymity due to the sensitive nature of the discussions, noted that the meeting was partially driven by concerns over Anthropic's latest AI model.
Anthropic has restricted access to its Claude Mythos system through what it calls Project Glasswing, currently available only to a limited group of security researchers and tech companies. The company's CEO recently visited the White House to discuss the technology with senior administration officials, including Treasury Secretary Scott Bessent, chief of staff Susie Wiles, and National Cyber Director Sean Cairncross.
Companies uncertain about government's intentions
A document obtained by POLITICO shows that the White House's questionnaire covers technical and policy matters, including which coding projects should receive priority attention and how public-private collaboration might function under initiatives like Project Glasswing. One question simply asked what the most effective federal government role would be.
Four sources told POLITICO that some industry representatives found the questions confusing, with several described as vague. Two sources said certain inquiries touched on internal security practices that companies felt uncomfortable disclosing without clear justification.
The questions sought information about what systems companies have already tested using AI, as well as their scanning and remediation priorities, according to the obtained document.
Legal battle with Anthropic continues
The White House has been working to resolve a protracted legal dispute with Anthropic over the company's attempts to impose ethical boundaries on government AI use, a conflict that prompted President Trump to ban federal agencies from using Anthropic's software in February. However, growing recognition of Mythos's cybersecurity capabilities, alongside fears that unauthorized parties may be exploiting the technology, has led multiple agencies to seek access to the tool.
The Department of War formally designated Anthropic as a supply chain risk in March, a decision the company is challenging in federal courts. Meanwhile, competitors like OpenAI have begun testing their own advanced cybersecurity-focused AI models, increasing pressure on the government to find a resolution.
Trump recently suggested during a CNBC appearance that relations with Anthropic may be improving, describing the company's executives as highly intelligent individuals who are "shaping up."
Source:Websites
