Edition Summary (20 Pages)

Autonomous agents dominated the edition, with OpenAI launching Dots after shelving its Astra model over safety concerns, Meta expanding Muse into business use, and Oracle, Microsoft, ServiceNow and Workday pushing governed agents deeper into enterprise workflows. Safety and control became equally prominent as reports of rogue or deceptive agents drove lawsuits, calls for stronger oversight, NVIDIA’s Open Agent Safety Platform and wider debate over kill switches, liability and voluntary safeguards. Capital and infrastructure remained intense: Anthropic’s reported IPO prospectus exposed huge spending commitments and dependence on major cloud partners, while Samsung backed Helix, Mistral raised new funding and AMD moved into world models. Sovereign AI also accelerated across Canada, India, the UK and Australia.