Anthropic’s Olah tells Vatican AI cannot be left to Big Tech alone

TL;DR:

• Anthropic co-founder Chris Olah sat alongside Pope Leo XIV at the launch of Magnifica Humanitas and argued frontier AI development cannot be left to technology companies alone.
• Olah said there was “a real possibility” AI will displace human labour at very large scale, calling support for displaced workers “a moral imperative of historic proportions”.
• He was the only Big Tech representative invited to the Vatican event, signalling Anthropic’s deliberate alignment with external moral and political oversight rather than industry self-governance.

Speaking in Vatican City on Monday, Olah told the audience that “every frontier AI lab … operates inside a set of incentives and constraints that can sometimes conflict with doing the right thing”, and that even well-intentioned researchers cannot fully escape those forces. The statement is striking from a co-founder of one of the world’s leading AI firms, and signals that Anthropic is prepared to publicly back arguments that the company itself cannot be trusted to self-regulate.

Context and Background

Anthropic was founded in 2021 by Olah and other former OpenAI staff who left over concerns the rival was moving too fast without thorough testing. The company has since clashed with the Trump administration over guardrails restricting how its Claude models can be used for autonomous weapons targeting and domestic surveillance — making the Vatican appearance a continuation of an established positioning strategy rather than a one-off.

In Anthropic’s full statement published the same day, Olah set out three areas requiring urgent external attention: the global poor and uneven distribution of AI gains, the question of human flourishing alongside increasingly capable systems, and the “mysterious, even unsettling” findings emerging from interpretability research on model internals. “AI development is concentrated in a handful of wealthy nations. How can we ensure the gains of AI are shared globally?” he asked.

For UK enterprise audiences, the message reframes a familiar debate. UK government policy has hesitated between mandatory AI regulation and lighter-touch sector-specific oversight; a leading frontier lab now publicly arguing that industry must not be allowed to set its own rules complicates the case for voluntary frameworks.

Looking Forward

Olah’s appearance suggests Anthropic intends to keep cultivating non-government, non-industry validators of AI governance — religious leaders, civil society, academia — alongside its existing regulatory engagement. For UK firms procuring frontier models, expect this positioning to shape Anthropic’s commercial messaging on safety and oversight throughout 2026, particularly into financial services, legal services and public-sector buyers who weigh governance posture heavily.