The European Parliament goes on the defensive: how neural networks are taking over lawmaking
The European Parliament has found an unconventional yet highly illustrative way to combat "AI slop" within the walls of the legislative body. Instead of a total ban on the use of generative neural networks, the institution is launching its own platform, EPGenAI Hub. This service will provide MEPs and staff with access to models from OpenAI, Meta, Anthropic, and France's Mistral. Thus, the leadership hopes to steer politicians away from public, uncontrolled websites toward a secure and regulated alternative.
Anxiety within the EU is growing by the day. AI-generated texts are infiltrating parliamentary work too quickly. MEPs and their assistants are increasingly using neural networks to draft speeches, inquiries, and amendments to laws. According to my data, around 2,100 people use neural networks daily—nearly a fifth of all staff. At a July meeting of several committee chairs, genuine alarm was sounded. Suspicion was reinforced by an anomalous surge in amendments requiring translation. The texts were characterized by a distinctly dry style and contained references to non-existent laws. These are classic signs of "AI slop"—low-quality, template-based content mass-generated without proper verification.
Irish MEP Barry Andrews, who chairs the development committee, acknowledged that many colleagues write inquiries and speeches using AI. In his view, this is a personal choice driven by the immense workload. However, for Andrews himself, such tasks remain key responsibilities of an MEP. For now, legislators are not required to disclose their use of neural networks.
European-style sovereignty
The new platform has already been presented to MEPs and staff during workshops. The official launch of the project is scheduled for September. The beta version allows users to choose between two types of neural networks: local models (Llama from Meta, GPT-OSS from OpenAI, and Mistral Small) and external services (ChatGPT and Claude Sonnet).
However, here lies the main paradox. By launching the platform, the parliament is expanding the presence of American AI giants. This comes amid Brussels' push to reduce technological dependence on the US. Previously, the institution replaced Google with the French search engine Qwant. Additionally, technicians disabled built-in AI on MEPs' tablets for cybersecurity reasons. The parliament's press service clarified that the list of available models may change in the coming months. The system is expected to include sovereign European solutions from new cloud providers. This is a strategically sound move, but it requires time and resources.
Similar stories are unfolding in Russia. MEPs were accused of using ChatGPT to draft laws. The lawmakers themselves deny working through AI. Journalists suspected neural network involvement in the preparation of draft law No. 1126815-8. The reason was a footnote in the explanatory note containing the tag source=chatgpt.com. Such "traces" typically remain when copying from a chat. Representatives of the State Duma explained the situation: a staff member used AI only to quickly find the original source. The data itself turned out to be entirely correct. The committee added that drafting laws requires deep expertise, and neural networks are not yet up to such a task.
My comment: The problem of "AI slop" in lawmaking is just the tip of the iceberg. While regulators try to control AI through their own platforms, the very nature of generative models remains vulnerable to hallucinations and manipulation. Without strict verification standards and mandatory disclosure of AI use, we risk getting not high-quality laws, but well-formatted yet empty content.