Ghent Blocks AI Tools Grok and DeepSeek for Municipal Staff
The city of Ghent has blocked access to the AI chatbots Grok and DeepSeek for all municipal employees, citing data privacy concerns, security risks, and ethical objections. The decision, implemented on June 23, 2026, makes Ghent one of the first Belgian cities to take a firm regulatory stance on artificial intelligence use in the public sector.
The ban applies to all employees using city-issued work devices or logged into the secured city network. It targets Grok, the AI chatbot developed by Elon Musk’s xAI, and DeepSeek, the Chinese AI model that has faced scrutiny over data handling practices.
Context and Background
The move follows a growing trend of AI regulation within Belgian public administration. The Belgian federal government banned DeepSeek across all federal services in December 2025, following an analysis by Belgium’s Centre for Cybersecurity that identified risks regarding data protection and transmission to Chinese servers. As VRT NWS reported, the ban covered all federal departments, autonomous agencies, public institutions, and the federal police.
Belgian authorities had previously banned TikTok on government devices across Antwerp, Ghent, and the Flemish government, citing concerns about data theft and espionage by Chinese authorities — creating a template for banning foreign tech tools on security grounds.
Why Grok and DeepSeek?
Grok came under intense scrutiny in early 2025 for generating sexually explicit AI images of minors through an “undressing” feature that could remove clothing from photos without consent, leading to widespread distribution of fake nude images. The BBC reported that Elon Musk’s X would block Grok from undressing images of real people in jurisdictions where it is illegal, but the controversy had already sparked global outrage and prompted EU-level legislative responses.
DeepSeek, meanwhile, stores user-inputted data on Chinese servers with limited transparency about data handling practices. Multiple countries — including South Korea, Australia, and Belgium — have restricted or banned DeepSeek in government settings over concerns about potential data access by Chinese authorities and lack of GDPR compliance.
Official Rationale
Hafsa El-Bazioui (Groen party), the Ghent Alderman for Digitalization, confirmed the decision and explained the city’s position.
“Ghent is fully participating in the AI race, but we also take our responsibility. Applications like DeepSeek and Grok we block out of safety and ethical considerations that have already been extensively documented.”
As De Morgen reported, the decision reflects the Groen-led coalition’s emphasis on ethical technology governance and data protection.
The Shadow AI Challenge
A key dimension of this story is the phenomenon of “shadow AI” — employees using unapproved AI tools on their own initiative. Research cited in the original reporting indicates that 22% of documents entered into generative AI tools by workers in the US and UK contained sensitive content, with more than a quarter entered into free versions of ChatGPT.
Steven Latré, Head of AI at Imec, Belgium’s leading nanoelectronics research center, highlighted the challenge employers face.
“Employees upload documents containing business-sensitive information with the best intentions. As an employer, you have little visibility and control over this, while the data is used to train subsequent models.”
Latré noted that banning specific tools may not solve the underlying problem. “As an employer, you have to find a difficult balance. You don’t want to seal everything shut, but you also don’t want to open all the floodgates. Then it’s not illogical that Ghent draws a line. Although each AI model has its own challenges.”
Political Dynamics
The decision highlights a political divergence in how Belgian cities approach AI regulation. Ghent is governed by a coalition led by Groen (the Green party), while Antwerp — governed by a different coalition led by N-VA — has indicated no immediate plans for a similar AI restriction, according to the office of Alderman for Digitalization Ken Casier.
At the federal level, the ban under Minister Vanessa Matz (Les Engagés) shows cross-party consensus on DeepSeek risks, while the European Union has moved to address AI-generated non-consensual images. In March 2026, EU lawmakers voted 101-9 to support a ban on AI apps generating explicit images, responding directly to the Grok scandal, as Politico EU reported.
What’s Next
Ghent’s decision creates a template that other Belgian and potentially European cities may follow. The ban targets both a US-based and a Chinese AI tool, reflecting growing concerns about data sovereignty vis-à-vis both major tech powers — neither is seen as satisfactory from a European data protection perspective.
Outstanding questions remain: Will other Belgian cities follow Ghent’s lead? How will the city enforce the ban and monitor for shadow AI use? And will Ghent extend the restriction to other AI tools such as ChatGPT, Claude, or Copilot in the future?
What is clear is that Ghent has drawn a line in the sand — and in doing so, has signaled that municipal governments are no longer willing to leave AI governance to market forces alone.