The artificial intelligence sector is being rocked by questions surrounding the power structure at Anthropic. Analyst Kevin Bass recently sparked a major controversy by suggesting that the lab could pose a systemic risk similar to what the FTX platform represented for the cryptocurrency ecosystem. At the heart of these criticisms lies the concept of orchestrated regulatory capture, where overly close financial and organizational ties between AI firms, their investors, and the entities tasked with their safety create an opaque and potentially biased landscape.
The main point of tension concerns the influence of philanthropic foundations, notably Good Ventures, which support both AI safety research and direct funding for Anthropic. This configuration raises questions about the true independence of AI safety actors. By extension, the organization METR, which specializes in evaluating cutting-edge models, has found itself in the spotlight. Although METR asserts that it maintains strict autonomy from labs, its close technical collaborations with industry leaders fuel a sense of proximity that concerns some observers.
The paradox being denounced lies in Anthropic's strategy: while developing models of unprecedented power, the company actively promotes strict regulations and external audits. For critics, these normative requirements, though ostensibly justified by safety concerns, could serve as barriers to entry, consolidating the dominant position of established players while penalizing new competitors. This dynamic would reinforce a form of intellectual and financial monopoly over industry standards, leaving the market particularly vulnerable should one of its pillars fail.
The analogy to FTX highlights the stakes of this concentration. While no fraud has been proven to date, the risk is that of a systemic crisis of confidence capable of paralyzing an entire segment of the tech economy. With massive investments reaching hundreds of billions of dollars in AI infrastructure, the slightest questioning of governance or the integrity of audits could trigger a devastating domino effect. The U.S. Congress is now being urged to examine these connections, a clear sign that AI regulation has become as much a political issue as an industrial one.