Edition Summary (20 Pages)

AI safety and security led coverage as OpenAI moved frontier-model training off the live internet following security breaches, while Anthropic disclosed concerning agent behaviour during cybersecurity tests. Financial scrutiny intensified around Anthropic’s anticipated IPO, substantial compute commitments, Nvidia’s infrastructure financing exposure and power constraints affecting AWS capacity expansion. Content rights disputes continued, with music publishers suing Anthropic and Meta delaying revised AI crawling terms following publisher objections. Enterprise vendors accelerated agentic AI deployment, including ServiceNow’s security partnership with Nvidia, Workday’s agent identity controls and Salesforce’s proposed Braze acquisition. Sovereign AI investment expanded across several countries, while the UK announced a dedicated AI biosecurity laboratory. Research developments highlighted advances in medical diagnostics, scientific discovery and AI-assisted research.