Top themes today: Gemini 3.5 & Gemini Omni launch, OpenAI model roadmap, $33B+ AI startup funding, Google TPU at Next '26, four-state US AI law regime, prompt injection security standard.
AIGoogle's May 2026 model pair brings advanced agentic reasoning and multimodal creation to the Gemini API and AI Studio.
AI ResearchNew architecture and pipeline changes hold out the prospect of significantly cheaper frontier-model training.
Applied AILeading home hubs now bundle local voice, sensors and on-device routines, reducing cloud round-trips for everyday automation.
CloudMajor providers are expanding compute clusters, lowering inference costs and tightening enterprise tiers going into H2 2026.
CloudThe largest-ever Asia commitment for Microsoft covers AI and cloud infrastructure build-out through 2026-2029.
ChipsNew tensor processing units power live Google Cloud AI services, targeting training and inference workloads at scale.
ChipsAutomotive, connectivity and data-centre chip families span the full stack from end-device inference to cloud training nodes.
StartupsCrunchbase confirms 2026 AI venture already exceeds 2025 total; Together AI ($800M) and TwelveLabs ($100M) close major rounds on 1 July.
PolicyA June 2026 EO coordinates national AI policy, balancing capability growth with security and national-interest reviews.
PolicyFour states have active AI rules; FTC is actively fining companies — compliance programmes are now a board-level priority.
SecuritySecurity teams adopt layered guardrails as prompt-injection attacks grow in scale, sophistication and enterprise targeting.