Why AI Leaders Are Calling for External Regulation

In a rare alignment of interests, leaders from top AI firms like Anthropic and the Vatican have issued a joint call for robust, external regulation of artificial intelligence. Industry executives, including Anthropic’s Dario Amodei, have candidly admitted that the competitive pressures of the tech sector prevent companies from effectively self-regulating. Despite internal safety frameworks and red-teaming efforts, these measures remain siloed and voluntary, leaving the industry vulnerable to a collective action failure where unilateral restraint is viewed as a strategic disadvantage.

The core issue is that current AI governance is fragmented and internal, failing to address the systemic risks posed by autonomous systems, labor disruption, and mass surveillance. Because companies operate in a high-stakes, competitive environment, they lack the individual incentive to prioritize safety at the expense of market share. This creates a governance architecture problem rather than a purely technical one, as no single firm or nation can unilaterally mitigate risks without risking competitive obsolescence.

To address this, the article argues that the global community should move away from the nuclear arms control analogy and instead look toward the financial sector’s regulatory model. By utilizing frameworks similar to the G20’s Financial Action Task Force (FATF), the industry could implement shared norms, operational standards, and distributed accountability. This approach would integrate both governmental and private-sector obligations, creating a system where non-compliance carries real consequences. By focusing on high-level outcomes rather than rigid technical specifications, a global governance structure could effectively manage AI risks without stifling the innovation necessary for the technology's evolution.