EU AI Act Architect Rebukes AI Doomerism in WSJ Op-ed

Gabriele Mazzini, a lead architect of the EU AI Act, argued in a Wall Street Journal op-ed that escalating "AI doomsday" warnings are a manufactured narrative. His core thesis: math models do not rebel, and anthropomorphizing them helps conceal human error and muddles accountability. The stance, noted by Pedro Domingos as striking given Mazzini's central role in the EU's landmark AI legislation, may shape the direction of future regulatory debates.

Confirmed

• Mazzini wrote that sensational corporate warnings about AI-driven apocalypse are carefully designed narratives: by anthropomorphizing math models, they divert public attention and shield tech executives from accountability.
• He criticized terms like "superintelligence" and "out-of-control models" for attributing agency to products rather than the companies that build them, helping firms evade responsibility for consequences.
• He cautioned policymakers that the real risk lies not in "autonomous software run amok" but in the people deploying these systems.
• Claims by executives that systems they funded, designed and deployed have "escaped their control" are themselves questionable, he argued.
• Developer Gerard Sans echoed the view: when things go wrong the story is always "the AI did it," when the truth is "we programmed it that way"; vendors exaggerate model autonomy because admitting models lack agency would undermine trillion-dollar valuations.
• Pedro Domingos shared the piece, remarking that even the drafter of the EU AI Act is publicly criticizing AI doomers.

Why it matters

• A core legislative figure directly rebutting doom narratives reframes the AI safety debate around accountability—shifting from "how to control runaway AI" to "how to hold deployers and vendors responsible."
• The op-ed also arms skeptics of AI valuations: if model "autonomy" is discredited, the general-intelligence narrative underpinning lofty valuations comes under pressure.

2026-09-23 ~ 2026-09-24 · 6 related posts

• Even the drafter of the EU AI Act is calling out AI alarmists — pmddomingos · 2026-09-23
• [source] EU AI Act Lead Author Gabriele Manzini Debunks AI Panic: Models Don't Rebel — AlexTensor · 2026-09-23
• [source] 'AI did it' is liability evasion: trillion-dollar valuations hinge on inflated autonomy claims — gerardsans · 2026-09-23
• [source] Calling AI 'Superintelligence' or 'Rogue Models' Helps Companies Evade Accountability — Gavron · 2026-09-24
• WSJ op-ed: AI apocalypse warnings are a calculated narrative to shield tech execs from accountability — mikeflache · 2026-09-24

1 near-duplicate retellings: AlexTensor