Marc Warner on the Anthropic shutdown and the limits of AI sovereignty
Writing in Fortune on 23 June 2026, Marc WarnerDealroom has a profile for this one. Try Dealroom → — co-founder and CEO of Faculty, and now CTO of Accenture following its acquisition of the company — argues that Anthropic's suspension of foreign access to its most advanced models exposed a dependency European executives had been reluctant to confront. His take:
The kill switch is no longer hypothetical. “The idea that the U.S. could, in principle, simply cut off access to these systems is one that almost nobody took seriously until this weekend,” Warner writes. When Washington ordered Anthropic to suspend foreign access to its Fable 5 and Mythos 5 models, organisations across Europe and beyond that had embedded the tools into their workflows found them gone overnight — with no appeal, migration window or warning.
A vulnerability executives avoided naming. The episode, Warner says, surfaced a strategic exposure many leaders had been “very reluctant” to acknowledge: critical operations running on frontier AI controlled by a single foreign provider, subject to another government's decisions.
Sovereignty is about capability, not just infrastructure. Consistent with his earlier writing, Warner frames true AI sovereignty less as a question of where data centres sit and more as one of who holds the engineers, researchers and applied know-how to actually build and run the systems — the capacity to operate without the tool if access is withdrawn.
The lesson for European leaders. Rather than abandoning AI, Warner urges executives to build resilience: map where AI is embedded in critical workflows, avoid single-provider dependence, demand contractual clarity on suspension and data retention, and retain enough in-house expertise to keep functioning if a model disappears.
Read more: Fortune