Topics in this news episode:

• The autonomous AI attack on Hugging Face — what really happened

• Why OpenAI’s cyber test environment became a problem

• Zero-day marketplaces explained in plain language

• The asymmetry between attacking and defending AI systems

• Why locally or sovereignly operated models matter for cyber defense

• Anthropic: Fable 5, Opus 5 and the evolving model landscape

• Soofi S: an open German AI model for industrial applications

• IPCEI-CIS and Europe’s path toward digital sovereignty

• OVH Cloud, StackIT and sovereign cloud as a real architecture decision

• AI-Native SAFe, AI governance and the risk of putting frameworks before practice