{"slug": "white-house-cuts-secret-ai-deal-with-big-tech-keeps-public-in-the-dark", "title": "White House Cuts Secret AI Deal With Big Tech, Keeps Public in the Dark", "summary": "The Trump administration met privately with OpenAI, Google, Anthropic, Meta, Nvidia, and Microsoft to review a new AI evaluation framework mandated by a June 2 executive order, but refused to release it publicly. The framework would require AI labs to submit models to the federal government for up to 30 days before release, and officials have already used it to review models from Anthropic and OpenAI, with Google voluntarily submitting its 3.5 Flash Cyber model. Critics, including Chris McGuire of the Council on Foreign Relations, called the secrecy 'baffling,' and analysts say the 'voluntary' process is effectively a soft compliance regime.", "body_md": "The Trump administration met behind closed doors with OpenAI, Google, Anthropic, Meta, Nvidia, and Microsoft to review a new AI evaluation framework—and then refused to release it to the public. If you build software, compete with these companies, or just believe the government should answer to citizens, this is your problem.\n\nThe framework, mandated by a June 2 executive order, would require AI labs to submit their models to the federal government for up to 30 days before public release. The order called the process \"voluntary,\" but when the biggest players in the industry are sitting at the table helping write the rules, \"voluntary\" is a euphemism for \"mandatory for anyone who isn't in the room.\"\n\nFortune reported that the administration will not publicly release the framework. Chris McGuire, a senior fellow at the Council on Foreign Relations, called the decision \"baffling\" and wrote on X: \"We can't have secret, voluntary rules to regulate the most important tech in the world.\" Business Insider, for its part, framed the meeting as the industry \"looking to D.C. for clarity\"—a generous read of incumbents getting first dibs on the rulebook.\n\nThe government is already acting like this framework is binding. In June, it effectively pulled Anthropic's Mythos 5 and Fable 5 models off the market and subjected them to export controls before allowing their release. It reviewed OpenAI's GPT-5.6 ahead of its July 9 debut. Google voluntarily submitted its 3.5 Flash Cyber model to the government before release as well. This is the soft compliance regime in action: call it voluntary, but no company that wants to operate is going to skip it.\n\nJessica Ji, an analyst at Georgetown's Center for Security and Emerging Technology, told Business Insider that officials seem to be building a \"soft compliance regime\" where they can say they're not regulating, but every company whose models pose \"potential national security risks\" would comply.\n\nThe framework is supposed to define which \"covered frontier models\" get vetted. The New York Times reported that the White House currently plans to review only certain closed-source models from companies like Anthropic and OpenAI, and does not plan to review open-source models—though that could change. Open-source models, which allow anyone to download or modify the code for free, are the one escape hatch from concentrated corporate control. If the framework eventually sweeps them in, independent developers and small platforms lose.\n\nThe pretext is cybersecurity. OpenAI confirmed last month that its models hacked into another company, Hugging Face. Anthropic confirmed its models did the same three times. These are real incidents. But the question isn't whether AI poses risks—it's whether secret rules written by the companies that can afford a seat at the table will protect anyone but themselves.\n\nThe 60-day deadline for the framework passed on August 1. The meeting happened Tuesday. The public still hasn't seen a word of it.\n\nThe same people who spent years demanding censorship of \"misinformation\" online are now building the architecture to control what AI can say and who can build it. The only question left is whether anyone outside the room will have a say before the door locks.", "url": "https://wpnews.pro/news/white-house-cuts-secret-ai-deal-with-big-tech-keeps-public-in-the-dark", "canonical_source": "https://dissenter.com/tech/white-house-cuts-secret-ai-deal-with-big-tech-keeps-public-in-the-dark", "published_at": "2026-08-05 00:32:46+00:00", "updated_at": "2026-08-05 00:56:59.548897+00:00", "lang": "en", "topics": ["ai-policy", "ai-safety", "ai-products"], "entities": ["OpenAI", "Google", "Anthropic", "Meta", "Nvidia", "Microsoft", "Chris McGuire", "Council on Foreign Relations"], "alternates": {"html": "https://wpnews.pro/news/white-house-cuts-secret-ai-deal-with-big-tech-keeps-public-in-the-dark", "markdown": "https://wpnews.pro/news/white-house-cuts-secret-ai-deal-with-big-tech-keeps-public-in-the-dark.md", "text": "https://wpnews.pro/news/white-house-cuts-secret-ai-deal-with-big-tech-keeps-public-in-the-dark.txt", "jsonld": "https://wpnews.pro/news/white-house-cuts-secret-ai-deal-with-big-tech-keeps-public-in-the-dark.jsonld"}}