Trump administration enforces AI regulation through ad hoc interventions despite anti-regulatory stance

Trump Administration Enforces Shadow AI Regulation Through Anthropic Takedown

The Trump administration sent an export control directive to Anthropic nearly a week ago, forcing the AI lab to pull Claude Mythos and Fable 5 offline in a move that exposed the administration's approach to AI regulation 1

1

. Despite positioning itself as anti-regulatory, the White House is shaping the AI industry through ad hoc interventions without publishing clear standards or formal rulemaking processes 3

3

. After days of negotiations between Anthropic and the White House, the two parties remain at odds about how to restore access to the models, with Anthropic maintaining it violated no concrete procedures while officials contend the company behaved recklessly with frontier AI risks 1

1

.

Commerce Secretary Howard Lutnick's signature appeared on the letter that sparked the confrontation, imposing export controls that effectively created a licensing regime potentially impacting other AI labs 2

2

. The directive prohibited all foreign nationals from accessing the models, preventing many of Anthropic's own employees from using its cutting-edge technology and locking out major customers including Apple, Meta, and much of the Fortune 500 1

1

.

National Security Concerns Drive White House AI Policy Shift

US officials grew concerned earlier this month when they learned Anthropic shared Claude Mythos with SK Telecom, a South Korean telecom giant they allege has ties to China 1

1

. Separately, Amazon CEO Andy Jassy raised concerns to Treasury Secretary Scott Bessent that some guardrails on Fable 5 could be circumvented, highlighting ongoing challenges with AI model jailbreaking 1

1

. These national security concerns prompted the administration's intervention, though Anthropic says it coordinated with the US government on the Mythos rollout and immediately revoked SK Telecom's access when concerns were raised 1

1

.

A former White House technology official characterized the situation as a "slap-dash approach" resulting from the administration's extreme anti-regulatory posture colliding with real AI capabilities that experts have predicted for years 1

1

. The lack of systematic preparation and policies to manage benefits and risks has put the AI industry in a quandary as companies like OpenAI, Google, and Meta watch Anthropic's misfortune and wonder how to avoid similar fates 1

1

.

Power Dynamics Shift Within White House AI Leadership

With David Sacks stepping back from day-to-day involvement and Siriam Krishnan preparing to leave by the end of this month, influence over White House AI policy is shifting to a broader group of officials 2

2

. Howard Lutnick has emerged as a central figure, leading meetings on the sidelines of the G7 summit to discuss expanding access to advanced AI models and standing next to President Trump during summit press conferences 2

2

. One senior administration official said Lutnick is "fixing a problem" rather than "being a problem" following the export controls on AI letter 2

2

.

Scott Bessent has become a key player despite not directly overseeing AI testing, serving as the point of contact when Amazon raised concerns about Anthropic safety issues 2

2

. Sources view Bessent as "the more reasonable actor" and someone "trusted by the private sector and critical infrastructure operators as a sober actor" 2

2

. White House Chief of Staff Susie Wiles, though not typically involved in day-to-day AI policy, proved receptive to Bessent's concerns about how Mythos could impact the financial sector, helping reopen communication lines with the company 2

2

.

Voluntary Frameworks and Global AI Standards Emerge as Regulatory Tools

What has emerged is a shadow AI policy that shapes the industry's future without spelling out rules, operating outside formal rulemaking processes with few published standards 3

3

. Export controls, voluntary frameworks, and federal AI procurement guidelines are becoming the building blocks of the administration's approach to governing advanced AI models 3

3

. AI CEOs discussed with Trump at the G7 summit the possibility of what OpenAI's Chris Lehane described as a global forum for global AI standards 3

3

.

The administration's latest executive order established a voluntary framework for government review of some advanced AI models, and leading AI labs are working to understand compliance requirements 3

3

. The General Services Administration is also considering a new rule around safeguarding data when large language models process government information, setting privacy and cybersecurity standards for companies seeking government contracts 3

3

. At the G7 summit, the notion of tech sovereignty morphed large as foreign leaders grapple with not wanting to rely solely on American companies while recognizing they cannot ignore the world's leading models 3

3

.