The European Parliament sent an email to all members and their staff on Monday, announcing that it has disabled the built-in artificial intelligence functions on official equipment due to concerns about cybersecurity and data protection vulnerabilities. The email pointed out that some AI assistants will use "cloud services" to handle tasks that could be completed by local devices, which raises concerns about data leaks.
The email emphasized that as these features continue to evolve and expand to more devices, the scale of data shared with service providers is still being evaluated, and until the situation is fully clear, disabling such features is considered a safer approach.
In addition to official equipment, the European Parliament also urged lawmakers to take similar precautions on personal equipment, especially when these devices are used for work purposes. The email specifically warns that you should avoid letting AI functions scan work documents or private communications, and be extremely vigilant when installing third-party AI applications and granting access to their data.
In a statement to POLITICO, the European Parliament said its press service is continuously monitoring cybersecurity threats and rapidly deploying necessary preventive measures. However, Parliament declined to reveal exactly which built-in AI features had been disabled, citing the "sensitive nature" of the information, or what operating system the affected official equipment was running.
This is not the first such ban on AI tools targeting government equipment and systems. Previously, due to "national security" concerns, AI tools such as DeepSeek have been banned from use on government hardware in many countries and regions, including Taiwan, the United States, and Australia. In addition, the report also mentioned a security incident in August 2025, when Madhu Gottumukkala, acting director of the U.S. Cybersecurity and Infrastructure Security Agency (CISA), was found to have uploaded sensitive files marked "Official Use Only" to the public version of ChatGPT, an act that was detected by CISA's automated cybersecurity sensors and triggered an investigation by the Department of Homeland Security.