The US Congressional offices have been advised to avoid using DeepSeek, a chatbot developed in China that has been gaining momentum in the American AI landscape. Amid growing concerns over security and operational risks posed by evolving AI technologies, the House’s Chief Administrative Officer has issued precautionary guidance to mitigate potential threats.
According to a notice obtained by Axios, DeepSeek is currently under review and has not been authorised for official use within the House. The advisory warned of potential security risks, noting that cyber threats are already leveraging the chatbot to deploy malware and compromise devices.
To address these concerns, security measures have been enforced to limit DeepSeek’s accessibility on all House-issued devices. Congressional staff are strictly prohibited from installing or using the chatbot on government-provided phones, computers, and tablets.
This restriction follows previous AI-related limitations imposed by the Chief Administrative Officer. In 2023, the House allowed only the paid version of ChatGPT for specific tasks, and in April last year, staffers were barred from using Microsoft Copilot. However, Microsoft is reportedly developing AI tools tailored for government use, which may receive approval in the future. The decision to block DeepSeek reflects Congress’s ongoing efforts to balance AI innovation with the need to protect sensitive government systems from potential security threats.
Meanwhile, Italy’s data protection authority took steps to block access to DeepSeek. The regulator, known as Garante, cited concerns over user data protection and announced an investigation into the companies behind the chatbot. Garante expressed dissatisfaction with DeepSeek’s response to its inquiries regarding data collection practices, storage locations, and user notification mechanisms.
DeepSeek, founded in 2023 by Chinese entrepreneur Liang Wenfeng, has made a significant impact in the AI landscape. The company has developed advanced AI models capable of competing with industry leaders such as OpenAI and Anthropic
