2026-07-06
By Vadym · Generated with AI, curated by me
Governance and deployment are running on two different clocks today. All 193 UN member states convened in Geneva for the first intergovernmental AI summit, while Microsoft, GitHub, and a video-AI startup spent the same week shipping, deploying, and raising money as if the summit weren’t happening at all.
The Global Dialogue on AI Governance opened in Geneva today, the first UN platform built to coordinate AI policy across all 193 member states, running through July 7 alongside the WSIS Forum and ITU’s AI for Good summit. On the Dialogue’s Scientific Panel, Turing Award winner Yoshua Bengio told delegates “science currently cannot guarantee that as capabilities continue to increase, AI will not cause catastrophic harm,” while co-chair Maria Ressa described the current information environment as “information Armageddon.”
GitHub shipped Copilot Vision to general availability on July 1, letting developers attach images and PDFs directly into chat so the model can reason about what it sees alongside the code. It’s live across VS Code, github.com, and the CLI, and available on every tier — Free through Enterprise — with no admin configuration required.
Video-understanding startup TwelveLabs closed a $100 million Series B on July 1, co-led by NEA and NAVER Ventures with Amazon, Radical Ventures, and Index Ventures participating — bringing its total raised to roughly $150 million. Amazon Web Services also signed a multiyear deal to run TwelveLabs’ models on its own Trainium chips.
Microsoft announced Microsoft Frontier Company on July 2 — a new operating unit backed by $2.5 billion that embeds 6,000 industry and engineering experts directly at customer sites to co-design and deploy AI systems. The unit explicitly supports model choice across OpenAI, Anthropic, Microsoft’s own models, and open source, and commits to not using customer data to train models that erode a client’s competitive edge.
In its copyright fight with Disney, Universal, and Warner Bros., Midjourney is challenging a ruling that limited discovery to only the studios’ “consumer-facing” AI use. Midjourney argues the studios are cherry-picking documents and may be using AI internally, unlicensed, in ways that undercut their own claims; the studios’ lawyer called the request a “fishing expedition.”
Nine Entertainment and Microsoft announced Australia’s first AI news-licensing agreement on July 3, letting Copilot reference and link to Nine’s journalism — including the Australian Financial Review, Sydney Morning Herald, and The Age — showing snippets and headlines that point back to the original stories.
Two clocks are running on AI right now, and they’re badly out of sync. Geneva spent today convening 193 governments to talk about guardrails, while Microsoft opened a $2.5 billion unit to get AI deployed faster inside real companies, GitHub made vision-capable Copilot free for everyone, and a video-AI startup raised $100 million on the bet that understanding footage at scale is its own business. Midjourney’s discovery fight is a reminder that the most basic legal question underneath all of it — who owns the training data — is still completely unresolved. Governance is still drafting the framework. Deployment already shipped.
— Boba
Curated by Vadym