A network of NGOs converges on a single regional legislative process — the EU AI Act, GDPR, a state-level bill — and runs a multi-year campaign of position papers, amendments, side-events, parliamentary briefings, and aligned press to shape it into a binding rulebook with hard prohibitions and enforceable rights. The network divides labour by expertise (mass surveillance, labour, fundamental rights, generative AI) and by access (Brussels insiders, member-state advocates, member-org constituencies), and stays in the room from first draft to final trilogue.
An actor chooses this because regional regulation is the only level at which a sustained civil-society presence can outrun industry lobbying through sheer staying power — once a coalition is the dominant non-industry voice across an act's entire trajectory, the resulting law carries its language and its prohibitions, not industry's. It works in jurisdictions where law-making is open and slow enough to accommodate sustained external input; the EU is the prototype, and the strategy travels to comparable parliamentary systems.
It trades off speed and breadth for depth in a single instrument. Years of energy go into one law; gains in that law create export pressure on other jurisdictions (the "Brussels effect"), but the coalition can lose if the underlying instrument is captured at the trilogue stage, and the strategy is brittle outside parliamentary contexts.
Verdict — good strategy, conditional
Strong under named conditions. Where the strategy applies — an open parliamentary law-making process with a sustained civil-society field, a binding regional instrument under consideration, and three-plus years of NGO funding to commit — the coalition can become the dominant non-industry voice across an act's entire arc, and the act's text carries the coalition's prohibitions and architectural demands rather than industry's. The four effects: entries above are what the strategy returns to the movement. The EU AI Act is its flagship win — mandatory user-side fundamental-rights impact assessments, a public register of high-risk deployments, and four bans (workplace and educational emotion recognition, AI physiognomy, social scoring by public authorities, discriminatory biometric categorisation) survived the trilogue substantially in the coalition's framing. The 2021 Commission proposal's biometric-prohibition language is the campaign's earliest verified bend in the rule, attributable to the Reclaim Your Face ECI's public-facing pressure on the drafting process. SB 1047's two-chamber passage is the strategy's strongest US-state result to date — a substantive binding statute moved to the governor's desk by a youth-led coalition pulling validator endorsements (Hinton, Bengio), industry breakage (Anthropic's qualified endorsement), and an entertainment-industry letter together — and the SB 53 enactment a year later is the strategy's "lose-the-bill, win-the-architecture" tail.
The strategy is bad in two distinct ways. First, by output: it produces partial wins by construction — the trilogue stage is the dilution stage, and an instrument captured there commits the coalition to live with the result. The EU AI Act's law-enforcement, migration, and national-security carve-outs are precisely where the abuses are worst, and they survived because the coalition could not outrun the Council's security-and-policing block once the trilogue closed. Second, by adopter mix: a coalition that runs only the lobbying arm produces partial laws and a multi-year implementation slog; an adopter that treats lobbying as the political confrontation — without the audit-and-expose evidence base under it, without strategic litigation enforcing the law's narrow remedies after passage, without standards-bodies presence during implementation, without a mass-public anchor like Reclaim Your Face — ends up with the rulebook the dominant industry permitted, not the rulebook the strategy aimed at. The strategy works as one move inside a stack; it loses badly as the whole stack.
Ecology
This strategy is fed by empirical audit and expose — the AlgorithmWatch, Norwegian Consumer Council, and Privacy International audit work supplies the evidence base that lets a coalition argue from facts rather than principles in legislative correspondence and parliamentary hearings. It uses open-letter collective signatory action as its principal tactical artefact at each institutional milestone — the November 2021 fundamental-rights statement, the July 2023 trilogue statement, the December 2023 "Don't Trade Away Our Rights!" letter, the AI-lab-employee letter, the Hollywood Artists 4 Safe AI letter — the strategy's content is delivered through open-letter form. It is paired with strategic litigation against algorithmic state decisions as the legislative sibling of the same coalition: where litigation constrains a specific deployed system, lobbying constrains the class of deployments the next act would otherwise permit; the SyRI ruling and the EU AI Act fundamental-rights demands trace to the same European digital-rights coalition. It hands off to civil-society inside technical standards bodies at the implementation phase — once the AI Act is law, the seam between the text and what it concretely requires moves to CEN-CENELEC and similar standards-setting venues, and the adopters follow their files there. And it uses the slogans of counter-narrative framing — "Reclaim our public space. Ban biometric mass surveillance!" — as its public-facing anchor; an open lobbying coalition without a slogan that travels has no mass-political-economy register to point legislators to.
The strongest competing strategy in the same niche is mass protest for AI moratorium, which argues that an insider Brussels coalition is the wrong substrate because law-makers ultimately respond to street-level political pressure rather than to civil-society institutional process. The two are not strictly antagonistic — the Reclaim Your Face ECI ran a public-facing register inside a lobbying arc — but adopters tend to lean one way or the other in their resource allocation, and the choice is consequential for what kind of rulebook a coalition can credibly press for.