European Parliament switches off AI features over data fears

Some AI functions can involve transferring data off the device to external servers, official warn

The European Parliament has temporarily disabled AI features on lawmakers' official devices, citing concerns that sensitive information could be sent outside its secure systems.

According to internal communications seen by Politico, the Parliament's IT services have switched off AI tools on corporate laptops, tablets and other equipment after concluding they could not guarantee where the data might end up.

In an email to staff, the institution's technical support desk said some AI functions rely on cloud processing to complete tasks such as email summarisation or document drafting. That process, it warned, can involve transferring data off the device to external servers.

"Some of these features use cloud services to carry out tasks that could be handled locally, sending data off the device," the support team said.

"As these features continue to evolve, the full extent of data shared with service providers is still being assessed. Until this is fully clarified, it is considered safer to keep such features disabled."

Cloud concerns

The move reflects growing unease among European officials about the handling of sensitive information by AI systems, many of which are operated by companies based outside the EU.

Tools such as ChatGPT, Copilot and Claude typically process user requests in remote datacentres.

In certain jurisdictions, companies running these services may be required to provide user data to government authorities under national security or law-enforcement requests.

Officials are also wary that information uploaded to AI assistants could be used to improve the underlying models, raising the risk that confidential material might be exposed beyond its original context.

Studies in recent years have found that employees frequently paste internal documents, code or sensitive correspondence into AI tools, sometimes inadvertently breaching company policies or legal obligations.

Parliament officials stressed that the restrictions are temporary and targeted. Everyday applications such as calendars and email systems remain unaffected, and the ban is limited to AI features whose data-handling practices are not yet fully understood.

The decision is intended to remain in place until the IT department completes its assessment of what information is being shared and whether it can be processed entirely on the device rather than in the cloud.

Wider European debate

The Parliament's move comes as Europe continues to position itself as a global leader in AI regulation.

The bloc has enacted comprehensive legislation aimed at managing the risks of the technology. At the same time, the European Commission has proposed measures that could loosen certain data-protection restrictions to make it easier for companies to train AI systems on European data, a move critics say would benefit large technology firms.

The debate has intensified amid political tensions between the EU and the United States, where major AI providers are headquartered and subject to domestic laws.

In recent weeks, the US Department of Homeland Security has issued hundreds of subpoenas seeking information from major American technology and social media companies about individuals - including US citizens - who had publicly criticised the Trump administration's policies.

Meta, Google and Reddit reportedly complied in several instances, despite the requests not being approved by a judge or backed by a court order.