Where AI Value Lands

The Brussels Effect and the Brussels Mirage

4 min read

The Brussels Effect and the Brussels Mirage

Idea

The Brussels Effect is the phenomenon whereby the EU leverages its single-market scale to dictate global corporate behavior — non-EU firms standardize their global operations to EU compliance levels rather than managing fragmented regulatory regimes. The EU’s bet is that the AI Act will repeat this for AI. The Brussels Mirage, articulated in a 2026 LSE Law Review analysis, identifies the structural reason this bet is failing in AI: only the form of compliance travels abroad, not the substance of rights protection.

The Effect (as it worked for GDPR / DMA / DSA)

  • EU sets a stringent standard backed by access to ~450M consumers.
  • Multinationals comply globally because regional differentiation is operationally more expensive than universal EU-grade compliance.
  • The EU therefore exports norms without exporting jurisdiction — regulatory imperialism by market pull. This is the third of the three “digital empires” exports (digital-empires-tripartite).

The Mirage (as it is now failing for AI)

The LSE Brussels Mirage study draws a critical distinction:

  • De jure diffusion — formal extraterritorial application of EU law.
  • De facto diffusion — market-driven regulatory alignment.

De facto diffusion works well for consumer products (privacy, content moderation). It fails for high-risk, state-controlled domains like predictive policing.

Why. The EU AI Act is grounded in Article 114 TFEU — the internal market legal basis. It functions structurally as a product-safety instrument, not a fundamental-rights charter. What gets exported is therefore the compliance form: technical documentation, conformity assessments, impact-assessment procedures. What does not get exported is the rights-protective substance.

The Indian case study. Trinetra and CMAPS predictive-policing systems in India are deployed with EU-style technical documentation attached. State authorities point to this paperwork as evidence of due diligence — without conducting an independent Fundamental Rights Impact Assessment and without genuine human oversight. The compliance form legitimizes the deployment rather than constraining it. Instead of exporting fundamental rights, the EU has inadvertently exported a cover story that authoritarian or illiberal actors can use to wrap high-risk AI deployments in respectability.

Why this matters for value capture

If the Brussels Effect works for AI, the EU captures real value at the compliance-licensing layer — global enterprises pay an EU premium for trusted, auditable AI environments (divergent-value-stack-optima — regulatory-centric optimum).

If the Brussels Mirage is the truer picture, the EU’s regulatory bet exports paperwork without exporting power. Value capture collapses; the AI Act becomes purely a domestic compliance burden (sovereign-ai becomes the only remaining lever).

The handicap-or-moat question, sharpened

  • Short-to-medium term: handicap. Direct market evidence — the July 2025 open letter from 50+ European founders and VCs (Kry, 20VC, Voi, Synthesia) demanding implementation pause — confirms the AI Act is draining scarce VC capital toward legal overhead. The General-Purpose AI (GPAI) code of practice is cited specifically: mandatory external evaluations and stringent data-transparency requirements that US hyperscalers absorb as opex but EU startups cannot.
  • Long term: potentially a moat, but conditional. Only if (a) global enterprise/government AI demand matures around verifiable safety + auditability, and (b) the Brussels Mirage thesis is wrong, and (c) the EU domestic ecosystem survives the short-term capital drain. All three are contestable.

Graph View

Backlinks