This Week in AI: July 18 – 24, 2026
A federal judge granted final approval to Anthropic's landmark $1.5 billion copyright settlement with authors. OpenAI and Anthropic found rare common ground warning policymakers about open-weight Chinese AI models, Google shipped three new Gemini models, xAI brought Grok directly into Microsoft Office, and China's WAIC 2026 closed out with over 300 product debuts in Shanghai.
On this page+
- The Big Story: Judge Grants Final Approval to Anthropic's $1.5 Billion Copyright Settlement
- Source
- OpenAI and Anthropic Find Common Cause Against Open-Weight Chinese Models
- Source
- Google Ships Three New Gemini Models, Teases Gemini 4
- Source
- xAI Brings Grok Directly Into Microsoft Excel, Word, and PowerPoint
- Source
- WAIC 2026 Closes With Over 300 AI Product Debuts in Shanghai
- Source
- Strategic Takeaway
This week showed AI's biggest fights aren't just happening between labs anymore, they're happening in courtrooms, in Washington, inside the software people already use every day, and on exhibition floors in Shanghai. A copyright case that's been running for years finally reached a landmark resolution, two fierce competitors found common cause against a shared external threat, both Google and xAI pushed further into daily workflows, and China used its flagship AI conference to showcase how far its homegrown ecosystem has come.
The Big Story: Judge Grants Final Approval to Anthropic's $1.5 Billion Copyright Settlement
A federal judge in San Francisco granted final approval on July 20 to Anthropic's $1.5 billion settlement with a class of authors who accused the company of using pirated books to train Claude, according to TechCrunch. The payout works out to roughly $3,000 per work across an estimated 500,000 works, shared among the authors and publishers holding rights to them, per that same reporting. It's described as the largest known settlement of a US copyright case.
Source
OpenAI and Anthropic Find Common Cause Against Open-Weight Chinese Models
Despite competing fiercely for customers, OpenAI and Anthropic are aligning in Washington on warning policymakers about the risks of powerful open-weight Chinese AI models, according to Axios. The alignment follows Moonshot AI's Kimi K3, which launched the prior week and quickly topped several coding and agent benchmarks, intensifying industry debate about the competitive gap between US and Chinese labs. Per Axios's reporting, the two labs differ in their specific policy asks, but agree that open-weight models are harder to keep safe since developers lose the ability to revoke access or update safeguards once weights are public.
Source
Google Ships Three New Gemini Models, Teases Gemini 4
Google released three new Gemini models on July 21, Gemini 3.6 Flash, 3.5 Flash-Lite, and a security-focused 3.5 Flash Cyber built to find and patch software vulnerabilities, according to Google's own announcement and TechCrunch. Gemini 3.6 Flash reduces output token usage compared to its predecessor while improving coding and multimodal performance, per Google's release notes, and the company confirmed it has begun pretraining Gemini 4. The Flash Cyber model remains in a limited-access pilot restricted to governments and trusted partners.
Source
xAI Brings Grok Directly Into Microsoft Excel, Word, and PowerPoint
xAI launched free Grok add-ins for Microsoft Excel, Word, and PowerPoint on July 20, letting users interact with spreadsheets, documents, and presentations using conversational prompts, according to The Register and TechRadar. In Excel specifically, the add-in can answer questions about data, generate formulas, and run scenario analysis directly inside existing workbooks, per that reporting, putting Grok in direct competition with Microsoft's own Copilot inside Microsoft's own software.
Source
WAIC 2026 Closes With Over 300 AI Product Debuts in Shanghai
The exhibition floor of China's World AI Conference (WAIC) ran through July 20 in Shanghai, featuring more than 300 global product debuts spanning large language models, AI agents, humanoid robots, and industrial open-source ecosystems, according to Global Times and CGTN. Per that reporting, the largest crowds gathered around humanoid robots and the latest foundation models, with embodied intelligence, computing infrastructure, and AI governance emerging as recurring themes across the conference.
Source
Strategic Takeaway
The through-line this week is that AI's biggest battles are increasingly fought outside the model itself. Anthropic's settlement closes one of the industry's most consequential legal questions, whether training on pirated content carries real financial consequence, with about as decisive an answer as courts have given yet. OpenAI and Anthropic's alignment against open-weight Chinese models shows two bitter rivals recognizing a shared external threat matters more than their own rivalry, at least in front of policymakers, even as China's own WAIC showcase made clear that ecosystem is only getting more crowded. And xAI building its own presence directly inside Microsoft's flagship productivity apps, while Google keeps shipping cheaper, more capable Flash models, shows the distribution fight has moved from "build the best model" to "get inside the software people already have open all day."
