On the Vatican stage, Chris Olah from Anthropic states that AI cannot be guided solely by AI laboratories.

      During a presentation alongside Pope Leo XIV at the unveiling of *Magnifica humanitas*, the interpretability lead of a tech company acknowledged that incentives at frontier AI labs can sometimes lead researchers away from ethical practices. Christopher Olah, co-founder of Anthropic and head of its interpretability research, used the opportunity at the Vatican on Monday to assert a viewpoint not previously expressed by any major AI leader on such a significant platform: the advancement of frontier AI should not be solely entrusted to frontier AI laboratories. He spoke during the official launch of Pope Leo XIV’s first encyclical in the Vatican Synod Hall.

      “Every frontier AI lab," Olah stated, "functions within a set of incentives and constraints that can occasionally conflict with ethical considerations.” He emphasized that even well-meaning researchers are influenced by these dynamics. He concluded by stressing the importance of external oversight from religious leaders, governments, and civil-society organizations.

      The second part of his address focused on labor issues, where Olah warned of a “real possibility” that AI could displace human jobs “on a very large scale,” and if this occurs, it would be a moral duty of historic significance to support those affected.

      This statement marks the most explicit acknowledgment by a founder of a frontier AI lab regarding the potential rapid job displacement caused by the technology their company is developing. Anthropic's presence at the Vatican has quickly turned into the most prominent strategic shift of the year for any AI firm. The company previewed this new direction by announcing the opening of an office in Milan, positioning itself at the forefront of a significant statement from the Catholic Church regarding technology since Leo XIII’s *Rerum novarum* addressed industrial capitalism in 1891.

      Olah's role, overseeing the company's interpretability research, is regarded by Anthropic as its strongest argument for credibility concerning safety; he leads a team devoted to reverse-engineering the operations of frontier models.

      In contrast, the political context differs greatly from the moral one. This spring, Anthropic was at the center of two notable disputes with the US government. In April, the Pentagon removed the company from its top-secret AI projects due to its own usage limitations, opting instead for partnerships with Nvidia, Microsoft, and AWS. Additionally, the Trump administration halted the expansion of Mythos, an autonomous vulnerability-discovery model that has significantly impacted global bank cybersecurity governance. Olah’s call for external oversight, delivered alongside the pope, serves as a direct response to these challenges.

      This statement also comes at a crucial commercial juncture for Anthropic, as the firm is in discussions to secure $30 billion at a valuation of $900 billion. The contrast between the ethical and commercial pressures is striking, and Olah did not shy away from acknowledging this. He remarked that “companies like ours” face “strong commercial, geopolitical, and personal pressures that can conflict with the greater interests of society.”

      Olah argued that the solutions to these pressures must come from outside the lab. The encyclical outlines expectations for governments and civil society while Olah's speech provided a similar framework. Both effectively resisted the idea of placing the regulatory framework for the next decade in the hands of the companies that have been developing the technology it will oversee.

      The choice of Olah as the messenger was also deliberate. As the founder of an AI lab, he addressed an audience that included cardinals, the pope, and representatives from the White House, arguing that AI labs cannot tackle these challenges independently. Whether this argument translates into actionable policy remains uncertain, but the mere fact that a frontier-lab founder articulated it from within the Vatican is noteworthy in itself.