The European Parliament plans to launch its own platform that will allow members and legislative staff to use neural network tools in a secure environment, Politico reported, citing sources.
According to the newspaper, approximately 2,100 people, or roughly 20% of staff, use AI in their daily work at the European Parliament. Some lawmakers use publicly available neural networks to write speeches, interpellations and formulate legislative amendments. However, the European Parliament wants to create an official, unified platform and control the use of neural networks.
The European Parliament’s AI platform, called EPGenAI Hub, is scheduled to launch in September. The platform will allow MEPs and other European Parliament staff to have centralised access to models from Meta (designated extremist in Russia), Open AI, Anthropic and Mistral.
Users of the platform will be able to choose between neural network models running on the European Parliament’s own infrastructure and more powerful services hosted on external servers. The EP press service told the newspaper that the range of neural network models on offer could change in the future “given the emergence of open source models and models from EU sovereign states”.
In March 2024, MEPs approved a law on the use of artificial intelligence technologies. Under the legislation, companies creating “high-risk” AI systems must be more transparent about how they create and use AI models. Developers of the most powerful AI models face additional requirements, such as risk assessments. Companies that fail to comply with the law can face large fines or even be banned from the community altogether.