Loading post…
CloudflarePetra ArtsAnnouncement
One year later: Sovereign AI and the fight for choice
Summary
A year after arguing that AI sovereignty means choice, Cloudflare adds two European open models—EuroLLM and Apertus—to Workers AI, reflecting a year where access to frontier models became constrained and attacks using AI increased. EuroLLM supports 35 languages including all 24 EU official languages, many underserved by existing open models, and was trained on the MareNostrum 5 supercomputer. Apertus, Switzerland’s first fully open multilingual model, was trained on 15 trillion tokens across 1,500+ languages on the Alps supercomputer (10,000+ GH200 GPUs) and outperforms leading models on rare and regional languages. Cloudflare also launches hands-on workshops for government cyber agencies and critical infrastructure operators to build model-agnostic AI security defenses, starting in Singapore in October. The company open-sourced its own security harness that coordinates multiple AI models for vulnerability hunting, ensuring defenses remain operational even if access to one model is lost.
Context
Cloudflare argues that AI sovereignty is not zero-sum and that choice is key, reflecting on a year where access to frontier models became constrained and attacks using AI increased. The post presents Cloudflare’s approach of adding open models, building model-agnostic security tools, and running workshops to give nations genuine choice.
Approach / What changed
Adding EuroLLM and Apertus to Workers AI, launching hands-on workshops for government cyber agencies and critical infrastructure operators, and open-sourcing a model-agnostic security harness that coordinates multiple AI models for vulnerability hunting.
Takeaways
EuroLLM covers 35 languages including all 24 EU official languages, many underserved by existing open models, and was trained on the MareNostrum 5 supercomputer by a consortium of European universities and research institutions.
Apertus is Switzerland’s first fully open large-scale multilingual model, trained on over 15 trillion tokens across 1,500+ languages on the Alps supercomputer, and outperforms leading models on rare and regional languages.
Cloudflare open-sourced a security harness that coordinates multiple AI models for vulnerability hunting and verification, ensuring defenses remain operational even if access to one model is lost.
FAQ
What languages does EuroLLM support?
EuroLLM supports 35 languages, including all 24 official EU languages. Many of these languages are underserved by existing open models, and the model was trained on the MareNostrum 5 supercomputer by a consortium of European universities and research institutions.
What is Apertus and who built it?
Apertus is Switzerland's first large-scale, fully open, multilingual language model. It was developed by ETH Zurich, EPFL, and the Swiss National Supercomputing Centre (CSCS) as part of the Swiss AI Initiative, trained on the Alps supercomputer with over 10,000 GH200 GPUs.
How does the security harness work?
The harness is an orchestration layer that coordinates multiple AI models working in parallel to hunt for vulnerabilities, verify findings, and prioritize threats. It is model-agnostic, so losing access to one provider does not disable defenses, and Cloudflare has open-sourced it.
Source and freshness
- Source
- blog.cloudflare.com
- Published
- Added to Yomu
Summaries are generated from the original by a language model. Check the original before relying on it.
Keep reading
Related reading
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
The best engineering reads, once a week
One useful email. Unsubscribe anytime.