ENISA and the European Commission’s Joint Research Centre are quietly evaluating open-weight Chinese AI models to understand capabilities, limits, and risks.
Europe’s cyber and science agencies have started testing Chinese open-weight AI models as part of a broader risk assessment. The goal is to understand what these models can do, where they struggle, and what safety risks they pose. The tests run on the Commission’s secure systems, and officials say they are not being used in daily work. By year’s end, Europe plans to roll out an AI testing platform for sectors like energy and health. This move helps Europe decide how much it relies on foreign AI going forward and what safeguards to require for future adoption.
What this means for you is that, if Europe firms up safeguards, consumer-facing AI services could see tighter rules or different availability depending on where the tech comes from. In short, Europe is trying to balance access to powerful AI with safety and resilience.
The European Union is quietly expanding its look at foreign AI by starting to test open-weight Chinese language models. The EU’s cyber security agency ENISA and the Commission’s science service, the Joint Research Centre (JRC), have begun evaluating several Chinese models. Officials confirmed the move, noting that the work is not about using these models in day-to-day tasks but about understanding what they can do, where they hit their limits, and the hazards they might pose.
Open-weight models are released in full, meaning researchers can download, run, and study them. This is a key reason Europe is looking at Chinese options now, alongside its ongoing tests of American and European systems. The Commission says it has not asked for special access or licenses from the Chinese developers; the models are run on the Commission’s own secure computing environment.
In parallel, ENISA has told Politico that it is testing a small number of Chinese open-weight models while already having access to more mature systems from familiar players. The two bodies are also collaborating on an AI testing platform that would let critical sectors test tools on mock networks without risking real systems. The platform is expected by the end of this year.
The public safety angle here is part of a broader European push to understand foreign AI capabilities, balance risk, and decide how much to rely on external providers. Safety and security are not the only goals, but the tests are part of a careful look at what European infrastructure and regulation should require of AI tools from outside the bloc.
Officials say the tests reflect Europe’s broader strategy: understand what foreign AI can do, identify gaps, and prepare safeguards before any large-scale adoption. The outcome could influence how Europe commissions or licenses AI tools in critical sectors, from utilities to healthcare.
They are evaluating Chinese open-weight AI models, alongside existing American and European systems, to compare capabilities and limitations. The effort focuses on assessing what these models can accomplish, where they struggle, and the practical risks involved in using them.
Europe wants to balance access to powerful AI with safety and resilience. As tech leaders push for broader AI deployment, the EU is weighing how much it should rely on foreign providers and what guardrails are needed for critical functions.
For readers, the big takeaway is about choice and risk. If or when Europe decides to lean on foreign AI more heavily, it could shape which tools are approved for use in public services or contracted by private companies for sensitive work. It could also influence cost, availability, and privacy safeguards that apply to AI-powered services the average person encounters online.
The Commission says the AI testing platform should be ready by the end of the year, and more formal assessments or policy steps could follow depending on results. For now, the tests are about gathering evidence rather than delivering consumer products.
What models are being tested?
The EU is evaluating open-weight Chinese AI models, alongside its ongoing tests of American and European systems.
Will these be used in daily work?
No. Officials say the tests are for assessment and safety planning, not daily operations.
When will we know the results?
The testing platform is to be delivered by year’s end, with any policy steps to follow later depending on findings.
You're reading the quick version.
President Trump urges federal agencies to use the term ‘super intelligence’ instead of ‘artificial intelligence’ in official communications, a shift that could ripple through policy and tech.
The Altruists arrives Nov. 19 with eight episodes that recount Sam Bankman-Fried’s rise and the collapse of FTX.
The season opener uses a fictional AI-driven hit to spotlight safety, accountability, and the costs of innovation.
The Daily Brief
Every morning, the news that matters from AI, phones, apps and the people shaping tech. Explained in plain English. Free.
One email a day. No spam, unsubscribe anytime. Privacy policy
