AI Daily Digest — May 30, 2026

Coverage below reflects the Vancouver date of May 30, 2026.

1. Anthropic raises $65B Series H at a $900 billion valuation, surpassing OpenAI

Anthropic has closed a $65 billion Series H funding round that values the company at approximately $900 billion, according to The New York Times — exceeding OpenAI's last publicly reported valuation of $730 billion. Anthropic says the funds will go toward safety research, expanded compute, and scaling its products. The round marks a notable inflection: the safety-focused lab that positioned itself as a measured alternative to OpenAI has now outpaced it on paper. Observation: Reaching a $900B valuation on the back of a safety-focused brand signals that investors are treating Anthropic's positioning not as a competitive constraint but as a market differentiator — one that is pricing at a significant premium. Link: https://www.anthropic.com/news/series-h

2. OpenRouter closes $113M Series B as AI model-switching infrastructure draws major investment

OpenRouter, the API aggregation layer that lets developers route requests across dozens of AI models, has raised a $113 million Series B. The round reflects growing conviction that model-agnostic infrastructure — rather than any single model provider — is a durable business in a landscape where frontier models are being displaced on a near-monthly basis. Observation: A $113M raise for a routing layer signals that market participants expect model churn to continue long enough to make model-switching infrastructure a category unto itself, not just a transitional workaround. Link: https://openrouter.ai/announcements/series-b

3. OpenAI brings Codex computer use to Windows after Mac launch

Following its Mac rollout, OpenAI's Codex app now supports computer use on Windows, allowing the agent to see the user's screen and execute tasks on their device. Users can also monitor and manage Codex jobs remotely via the ChatGPT mobile app. The expansion accelerates OpenAI's push toward persistent, desktop-native agentic workflows across both major platforms. Observation: Shipping desktop computer use on both major platforms in quick succession is less about individual features and more about establishing Codex as a persistent, ambient agent layer that runs alongside a user's entire workday. Link: https://developers.openai.com/codex/app/computer-use

4. Microsoft consolidates GitHub Copilot, Copilot Chat, and Autopilot into a single AI super app

Microsoft is reportedly building an AI super app that combines GitHub Copilot, the Copilot chatbot, Copilot Cowork, and an internal agentic workflow capability codenamed Autopilot into one integrated product, Fortune reports. The move mirrors OpenAI's own super-app ambitions and reflects a broader platform consolidation play across the AI tooling landscape. Observation: Consolidating four separate AI surfaces into one product suggests Microsoft is trying to prevent fragmentation across its own AI portfolio before competitors can use that fragmentation as a wedge. Link: https://fortune.com/2026/05/29/microsoft-working-on-super-app/

5. OpenAI retires ChatGPT Canvas as GPT-5.5 Instant takes over

Canvas, the side-by-side code and text editing interface in ChatGPT, will no longer be available for GPT-5.5 Instant or GPT-5.5 Thinking. Subscribers can access Canvas through legacy models for a limited time. OpenAI is also trimming GPT-5.5 Instant's response length and reducing its reliance on bullet-heavy formatting to improve readability. Observation: Retiring Canvas as the default editing surface while simultaneously adjusting response formatting suggests OpenAI is converging on a single conversation interface for GPT-5.5 — treating feature surface area as a liability, not a selling point. Link: https://www.theverge.com/ai-artificial-intelligence

6. Illinois passes the strictest US AI safety law, mandating independent audits and whistleblower protections

Illinois Governor JB Pritzker plans to sign a state legislature bill requiring independent audits and whistleblower protections at AI companies. The bill goes further than recently enacted AI safety laws in New York and California, marking Illinois as the highest-bar regulatory jurisdiction in the US for AI governance so far. Observation: Independent audit requirements and whistleblower protections target the accountability gap that softer disclosure mandates leave open — Illinois is betting that external verification will produce better outcomes than self-reporting. Link: https://www.theverge.com/ai-artificial-intelligence/2026/5/28/illinois-ai-safety-law

7. Amazon shuts down internal AI usage leaderboard after employees game it with pointless tasks

Amazon has shuttered an internal leaderboard tracking which employees use AI the most, after discovering workers were assigning AI agents to meaningless tasks just to climb the rankings. A company exec told staff: "Don't use AI just for the sake of using AI." The episode is an unusually candid public admission that AI adoption metrics can produce perverse incentives. Observation: Gameable adoption metrics are a predictable failure mode of top-down AI mandates — measuring usage quantity rather than value produced creates pressure to generate busywork, not outcomes. Link: https://www.ft.com/content/b1a62a7f-6df5-4c90-94ce-64ce9c9961b6

8. ElevenLabs releases Music v2 with mid-track genre switching and commercially licensed training data

ElevenLabs has launched Music v2, an updated AI music generation model capable of switching genres mid-track — from opera to heavy metal, for example — while handling faster rap delivery and non-musical sound effects. The company says the model is trained exclusively on licensed data and cleared for commercial use, addressing a persistent IP concern in the AI audio space. Observation: Licensed training data is becoming a differentiator in AI audio tools as legal risk from unlicensed data starts to land on enterprises deploying outputs commercially. ElevenLabs is betting that provenance matters at the product level, not just the legal one. Link: https://elevenlabs.io/blog/music-v2