European Parliament Disables Device AI Features, Urges Nuanced Controls

The European Parliament disabled built-in AI features on corporate tablets and phones issued to MEPs and staff in February 2026. Per an internal email seen by Politico and reported by TechCrunch, the Parliament's IT department said it could not guarantee the security of data uploaded to AI company servers and that "it is considered safer to keep such features disabled." The restriction covers writing assistants, text-summarisation tools, enhanced virtual assistants, and webpage-summary features. Everyday tools such as email and calendars were unaffected.

In a June 2026 European Business Review article, Trevor Dearing argues the on/off framing misrepresents the actual challenge. He cites survey figures - 55% of security leaders view AI-powered attacks as a major risk, while only 19% rank unmanaged LLM use as a top concern - as evidence that the more dangerous gap may be internal, unsanctioned use rather than sanctioned built-in features. Dearing advocates for layered, adaptive controls: access restrictions, model governance, monitoring and data-loss prevention configured to the risk level of specific features rather than blanket prohibition.

The Center for Data Innovation published a March 2026 commentary by Matthew Kilcoyne making a parallel argument from a policy perspective. Kilcoyne notes that restricting built-in AI from corporate devices does not stop MEPs from using them - it pushes usage to personal devices and third-party apps outside institutional oversight, trading a manageable, auditable risk for an invisible one. He also highlights a practical cost: the Parliament operates across 24 official languages, and AI writing and translation assistants are productivity multipliers that smaller delegations with fewer staff can least afford to lose.

The February decision followed broader European concern about U.S. technology dependence at a time when several EU member states were reconsidering their reliance on American tech companies. The EP's IT email noted that cloud-hosted AI features mean data "could turn up on U.S. servers," where it falls under U.S. law and is potentially subject to government demands - a risk that had become more salient given the Trump administration's subpoena activity in early 2026.

Observers should track whether the Parliament revisits its policy (Kilcoyne notes the 2023 TikTok ban on staff devices remains in effect, suggesting restrictions can become permanent), how other EU institutions respond, and whether unsanctioned AI tool use on personal devices increases as a consequence.

Note: The European Business Review piece is an opinion commentary and the 55%/19% survey figures are cited by that article without attribution to a named study. They should be treated as the article's reported data, not independently verified statistics.