Trump's AI Crackdown: Who Wins as Administration Targets Anthropic?
Anthropic's latest AI models, Fable 5 and Mythos 5, have been taken offline due to a Trump administration export control order citing national security concerns, sparking widespread debate. The move, allegedly prompted by Amazon researchers bypassing model guardrails, highlights Anthropic's contentious relationship with the administration and raises questions about AI regulation and digital sovereignty. Despite the setback, some speculate this controversial event could paradoxically enhance Anthropic's public image and perceived power.
Anthropic recently removed its two newest AI models, Fable 5 and Mythos 5, from public and user access following an export control order from the Trump administration. This action has ignited extensive discussions regarding AI policy and digital sovereignty, prompting a detailed analysis on TechCrunch’s Equity podcast featuring Sean O’Kane, Rebecca Bellan, and Anthony Ha.
The immediate cause for the shutdown was a letter from the U.S. government, sent on a Friday evening, citing unspecified “national security concerns.” Anthropic was instructed to ensure that these models could not be used by foreign nationals, a mandate that led to the complete withdrawal of the models due to the impracticality of distinguishing foreign users from domestic ones, especially given Anthropic’s diverse employee base. Reports suggest that the White House was alerted to the issue by Amazon researchers who allegedly found a way to bypass Fable 5’s guardrails, with Amazon CEO Andy Jassy reportedly raising these concerns.
The incident highlights the strained relationship between Anthropic and the Trump administration, which has been notably more contentious than the administration’s interactions with other leading AI laboratories. This unique dynamic raises questions about whether similar crackdowns are imminent for Anthropic’s competitors or if this action is specifically targeted. While some might view it as an advantage for other companies to avoid similar entanglements, it also underscores a precarious regulatory environment where outcomes might be influenced by political relationships.
Adding to the complexity, independent security experts and cybersecurity researchers have heavily criticized the government’s decision. Many have signed an open letter urging the revocation of the order, arguing that it is counterproductive and dangerous to strip U.S. network defenders of advanced cybersecurity capabilities provided by these models. Anthropic itself stated that similar jailbreaks could be found in other AI models, leading to speculation that the move might be retaliatory, given a prior lawsuit and the government labeling Anthropic a supply chain risk.
Critics also point to a perceived hypocrisy within Anthropic’s public stance. The company had previously advocated for slowing down AI development due to safety concerns, only to release its “most insane ever, super powerful model” shortly thereafter. This behavior reflects a broader industry trend where leaders like Sam Altman and Jensen Huang discuss the dangers of AI while simultaneously promoting increasingly powerful systems, fostering public unease about AI's potential to displace jobs and create “God machines.”
Despite the challenges and evident stress for Anthropic, there's a cynical perspective suggesting this episode could inadvertently benefit the company. Previous confrontations between Anthropic and the Trump administration resulted in increased downloads for Claude, positioning it as a more “responsible” or “resistance” alternative to ChatGPT. This latest incident, by making Anthropic’s models seem even more potent and controversial, could again boost their appeal, playing into the “everyone loves a bad boy” narrative.