The European Parliament just disabled built-in AI features on all lawmakers’ and staff devices — tablets, phones, the works. Writing assistants, text summarizers, virtual assistants: all gone.
Not because of the AI Act. Not because of some sophisticated compliance framework they spent years building. Because their IT department sent an internal email saying they “cannot guarantee the security of the data.”
Let me get this straight.
The institution that spent four years crafting the most comprehensive AI regulation on the planet — 892 pages of risk categories, conformity assessments, and compliance obligations — looked at a text summarizer on an iPad and said: “Yeah, just turn it all off.”
They didn’t deploy a secure European alternative. They didn’t set up sandboxed environments. They didn’t even finish assessing what data these tools actually share. They just… hit the kill switch. The email literally says the “full extent of data shared with service providers is still being assessed.”
Still being assessed. In February 2026. After the AI Act entered into force.
This is the same Parliament that banned TikTok on staff devices in 2023. The same one where lawmakers are now pushing to ditch Microsoft entirely in favor of European alternatives that mostly don’t exist yet at comparable quality. See the pattern? Ban, disable, restrict, replace-with-something-that-doesn’t-exist-yet. Repeat.
Meanwhile, as I wrote ten days ago in Part 4 of my series, Anthropic shipped Opus 4.6, OpenAI launched a desktop Codex app managing fleets of coding agents, DeepSeek and Moonshot keep churning out frontier-class open-source models from China, and Arcee AI — a 30-person startup — trained a 400B model from scratch for $20M. The gap between “research preview” and “deployed in production” has collapsed to days.
And the European Parliament’s response to this moment? Disabling a writing assistant on a tablet. Because the IT department is “still assessing.”
This is exactly what I’ve been tracking in my“AI Regulation or AI Requiem”series. Part 1 laid out how the EU and US were diverging into two incompatible visions. Part 2 showed the Commission already panicking and watering down its own rules. Part 3 documented Christine Lagarde openly admitting Europe had missed the AI wave. Part 4 — just ten days ago — described an EU fighting itself while the rest of the world ships.
And now, Part 5 just wrote itself.
The US ships AI features. China ships AI features. The EU Parliament ships internal memos about disabling AI features and sends its lawmakers back to typing emails manually, like it’s 2019.
At some point, “precautionary principle” stops being a policy and starts being an epitaph.
The rest of the world isn’t waiting. And every month that Europe spends “still assessing,” the gap widens.
📚 The “AI Regulation or AI Requiem” series
Part 1: AI Regulation or AI Requiem?— Where it all started: two fundamentally different visions for AI, and why one of them is going to fail hard.
Part 2: Europe Blinks First— The Commission’s “Digital Omnibus” proposal: not a reform, but a confession of failure.
Part 3: Europe’s Complete Capitulation— Christine Lagarde admits Europe missed the AI wave. The woman who controls monetary policy is now begging to have the obstacles her peers built removed.
Part 4: The EU is fighting itself. The rest of the world is shipping.— The Omnibus farce continues while frontier AI moves at breakneck speed.
📰 Sources for this post
Politico — EU Parliament blocks AI tools over cyber, privacy fears— The original report, based on the internal email sent to lawmakers.
TechCrunch — European Parliament blocks AI on lawmakers’ devices— How cloud-based AI processing creates data sovereignty headaches for EU institutions.
eWeek — European Parliament Blocks AI on Lawmakers’ Devices— The stat that says it all: 33% of EU citizens used generative AI in 2025, but their own Parliament bans it.
TechRadar — EU Parliament bans AI use on government work devices— The broader context, including the push to replace Microsoft with European alternatives.