AI regulation is a mess, and Anthropic is caught in the crosshairs

AI Regulation in the U.S. Remains Fragmented, with Anthropic at the Center of Recent Disputes

AI regulation is a mess and Anthropic - Just days after its launch, Anthropic's advanced public AI model was swiftly removed from user access. The Trump administration cited a potential jailbreak—where the model could bypass its own security measures—as a justification for labeling it a national security risk. This led to an export restriction, limiting even Anthropic employees’ ability to use the system. However, the company argues the flaw does not justify such a sweeping measure. The disagreement underscores the chaotic nature of U.S. AI oversight, with experts acknowledging the need for government involvement in AI safety but questioning the lack of clarity in regulatory processes.

The Anthropic Controversy

Anthropic, a fast-growing AI lab valued at nearly a trillion dollars and preparing for a public market debut, has become a focal point in the ongoing debate over AI governance. The situation escalated when the Department of Defense added the firm to its supply chain risk list, citing concerns over its AI systems. Shortly after, Anthropic unveiled its latest model, Mythos, which sparked widespread cybersecurity debates due to its ability to identify vulnerabilities. The public version, Fable 5, was released on June 9, but the administration claimed some of its safeguards had failed, potentially giving hackers a powerful new tool.

In response, Anthropic halted access to both Mythos and Fable to align with government directives. This move has prompted calls for greater transparency in how cybersecurity decisions are made, with critics warning that the absence of a structured framework could hinder U.S. progress in the global AI race.

"The problem is not that the government exercised discretion; national security demands such latitude," Jessica Tillipman, associate dean for government procurement law at George Washington University, wrote in an essay. "What is striking is the absence of any meaningful process."

Anthropic and the administration have been in talks to address the issue, with Trump stating at the G7 summit that negotiations are "going fine." In a recent interview with Axios, he acknowledged a shift in perspective, saying he no longer sees the company as a national security threat. "Well, not now, but a week ago, maybe," he added, highlighting the evolving stance.

Meanwhile, the Trump administration has adopted a more lenient approach to AI regulation, favoring voluntary compliance over strict mandates. This includes rolling back Biden-era policies, such as mandatory safety reporting, and prioritizing state law preemption. In March, Trump released a national policy blueprint urging Congress to regulate AI through specialized industry entities rather than a single oversight body.

Despite these efforts, some states have taken independent action. California passed legislation requiring AI firms to disclose risk frameworks, report safety issues, and protect whistleblowers. Florida, on the other hand, launched a criminal probe into OpenAI and filed a lawsuit, alleging ChatGPT played a role in a campus shooting last year. OpenAI has disputed these claims, emphasizing its commitment to safety through multiple guardrails and initiatives.