Cloudflare Adds EuroLLM and Apertus to Workers AI, Announces AI Security Workshops
Summary
Cloudflare says EuroLLM and Apertus are coming to Workers AI and is launching workshops for government and critical-infrastructure teams to build model-agnostic AI defenses, including a multi-model harness for vulnerability hunting, verification and prioritization.
Key points
- Cloudflare says EuroLLM and Apertus will be available on Workers AI, with access available by request.
- EuroLLM supports 35 languages, including all 24 official EU languages; its consortium reports strong results on multilingual benchmarks and machine translation.
- Apertus is a fully open multilingual model trained on more than 1,500 languages; its developers say its architecture, weights, training data and methods are published.
- Cloudflare describes a harness that coordinates multiple AI models to hunt for vulnerabilities, verify findings and prioritize threats. The company says it open-sourced the harness for organizations to use with models of their choice.
- Cloudflare says it developed layered defenses intended to prevent attackers using frontier models from finding vulnerabilities.
- The new hands-on workshops are for government cybersecurity agencies and critical-infrastructure operators; participants will build an AI security harness and layered defenses. The first is planned for Singapore in October.
Article Details
- Event Type
- Cloudflare announced that EuroLLM and Apertus are coming to Workers AI, launched hands-on AI security workshops for government cyber agencies and critical infrastructure operators, and described an open-sourced, multi-model security harness.
- Impact
- The harness is designed to coordinate multiple AI models to find vulnerabilities, verify findings, and prioritize threats, while allowing organizations to choose models without depending on a single provider. The workshops will train participants to build and adapt AI security defenses. The article reports no quantified security outcomes or specific cyber incident.
Vendors
Products
ApertusThis year, two of them come from Europe: EuroLLM, which covers all 24 official EU languages, and Apertus, Switzerland's fully open model, trained on more than 1,500 languages. Both were built by publicEuroLLMWeek, when we traditionally ship presents to the Internet. This year, two of them come from Europe: EuroLLM, which covers all 24 official EU languages, and Apertus, Switzerland's fully open model, trained on moreWorkers AI1,500 languages. Both were built by public universities and research institutions. Both are coming to Workers AI, and you can request access today.
Countries
IndiaWe think the past year can point the other way. India, Japan and Singapore focused on open-sourced models, and people across Asia-Pacific built tools on them for rural citizens, elderly patients and the nurses who careJapanWe think the past year can point the other way. India, Japan and Singapore focused on open-sourced models, and people across Asia-Pacific built tools on them for rural citizens, elderly patients and the nurses who careSingaporeWe're also launching hands-on workshops that help government cyber agencies and critical infrastructure operators build AI defenses that work with any model. The first runs in Singapore in October.Switzerlandyear, two of them come from Europe: EuroLLM, which covers all 24 official EU languages, and Apertus, Switzerland's fully open model, trained on more than 1,500 languages. Both were built by public universities and
Industries
Critical infrastructureWe're also launching hands-on workshops that help government cyber agencies and critical infrastructure operators build AI defenses that work with any model. The first runs in Singapore in October.CybersecuritySo today we're launching a program of hands-on workshops for government cybersecurity agencies and critical infrastructure operators. Participants build their own AI security harness and layered defenses, and leaveGovernmentWe're also launching hands-on workshops that help government cyber agencies and critical infrastructure operators build AI defenses that work with any model. The first runs in Singapore in October.