OpenAI and Anthropic models receive government approval before release
OpenAI released GPT-5.6 to the public after the U.S. Department of Commerce approved a broad launch following government testing, marking the first model to transition from restricted release to full deployment under the Trump administration's new frontier AI oversight framework.
Objective Facts
ChatGPT maker OpenAI said its latest powerful artificial intelligence model series will be released to the public Thursday, as the U.S. government reportedly approved a broader launch. By July 8, it had received formal approval from regulators after the Commerce Department's Center for AI Standards and Innovation conducted additional testing, with OpenAI sending technical experts to Washington to address questions. The company's new GPT-5.6 offerings have drawn concern over their ability to identify weaknesses in code that hackers can exploit, with the series offering three tiers: Sol (flagship), Terra (midrange), and Luna (fast, low-cost). Anthropic said it would begin restoring access to its most powerful AI models, Fable 5 and Mythos 5, after Washington lifted a restriction on where they could be released, while the government is drawing up criteria for which AI models would fall under new security restrictions.
Left-Leaning Perspective
U.S. Rep. Lori Trahan, a Massachusetts Democrat and co-author of a bipartisan bill that would regulate AI, said in a statement that she is concerned "the Trump administration is deciding company by company who gets access to the newest AI model. No law. No process. No oversight. Just appointees in Washington deciding who's in and who's out." The Trump administration is requiring both Anthropic and OpenAI to get approval for each new customer of their most powerful AI technology. There have been deep divisions within the White House on how to handle the releases of these increasingly powerful models, with some officials insisting on a soft touch from government, while others have expressed concerns about their potential impact on security.
Right-Leaning Perspective
Dean Ball, a senior fellow at the Foundation for American Innovation and soon-to-be OpenAI employee, wrote on social media that "in a matter of weeks, U.S. federal AI policy has gone from implausibly libertarian to increasingly draconian and opaque," and stated that Trump's executive order was "really establishing a de facto involuntary licensing/preapproval regime for frontier models." Ball argues the problem compounds when the government doesn't have clearly defined safety standards, which could lead to endless launch delays that might not only give a hand to China in the AI race, but also jeopardize the billions of dollars going to AI infrastructure buildouts. Commerce Secretary Howard Lutnick defended the government's role, writing that "we have worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the US Government and strengthen America's leadership in AI," while Anthropic had launched both models promoting them as state-of-the-art across benchmarks before the export control took effect.
Deep Dive
The government approval of OpenAI's GPT-5.6 represents a watershed moment in U.S. AI policy—the first model to successfully transition from government-restricted preview to public release under Trump's June 2 executive order framework. The context is critical: in mid-June, Anthropic's Fable 5 and Mythos 5 were suddenly pulled from public access via export control directives citing cybersecurity risks; these restrictions were later partially lifted, but Fable 5 remains unavailable for general use weeks later. OpenAI's preview began June 26 under government request and was approved for broader release by July 8 after testing by the Commerce Department's Center for AI Standards and Innovation. This sequence creates the story's fundamental tension: Trump signed an executive order explicitly rejecting "mandatory governmental licensing, preclearance, or permitting," yet both companies are now operating under de-facto approval processes with no published criteria, no statutory process, and no appeals mechanism—making the framework "voluntary" only in name. Each side accurately diagnoses what happened but interprets it through fundamentally different governance philosophies. Libertarian-right critics like Dean Ball and cybersecurity experts like Alex Stamos argue the process is dangerous both substantively and procedurally: the government lacks the expertise to evaluate frontier AI safely, has not articulated what risks it fears, and is making approval decisions through classified channels controlled by 15-20 improvising officials. Stamos explicitly stated there is no factual basis for the government's restrictions on Fable 5, arguing that "if the administration is honest about wanting the United States to beat China in this race, this is the dumbest thing they could possibly do." Left-leaning critics, including Rep. Lori Trahan, emphasize the absence of transparent rules, warning that the government is deciding "company by company who gets access" with "no law, no process, no oversight." Progressive outlets flag the commercial dynamics—OpenAI's equity offer to the government, Sam Altman's political donations—as raising capture concerns. National security officials and Trump allies counter that the restrictions are justified: Anthropic itself warned in April that Mythos posed unprecedented cybersecurity risks, and David Sacks, co-leading Trump's AI/crypto council, blamed Anthropic CEO Dario Amodei for "spiking cortisol levels" by calling Mythos a "cyber weapon." The government's lighter touch with OpenAI, in this view, reflects OpenAI's better cooperation and clearer safety assurances during testing. What both sides miss or downplay: the approval divergence creates perverse incentives. OpenAI's rapid transition from restricted to public release sets a market precedent that cooperation with government oversight yields faster approval, while Anthropic's extended restrictions create punishment for resistance (the company refused Pentagon demands for unrestricted military access and was designated a "supply chain risk"). This is not necessarily evidence of favoritism—OpenAI's demonstrated prior disclosure, phased release acceptance, and safety testing are material differences—but it establishes a powerful signaling effect: frontier AI labs that negotiate successfully with government enjoy faster routes to market, while those that resist face extended delays and commercial damage. The unintended geopolitical consequence is severe: OpenAI and Anthropic's access restrictions are driving enterprise adoption toward open-source Chinese models like DeepSeek's GLM-5.2 (free, downloadable, fine-tunable on private servers), which the government cannot gate or control. Usage data show the three major U.S. frontier labs' combined market share dropped from 55% to 33% in six months. The irony is that Trump administration policy—intended to secure U.S. dominance in the AI race against China—may be accelerating the very concentration of Chinese open-source AI adoption it seeks to prevent.