1. The White House, not OpenAI, decided when GPT-5.6 ships The Trump administration asked OpenAI to stagger the release of GPT-5.6, its next flagship model, citing concern over potential security issues, according to The Information. OpenAI agreed.

2. Ford topped JD Power's quality ranking by rehiring the engineers its automation replaced Ford leaned on automated systems across production and design, betting on fewer defects at lower cost. The systems turned out to be less reliable than the company had assumed. Quality slipped.

3. Retail's biggest AI shift is the one shoppers never see The features customers notice, like virtual try-ons and chatbot stylists, are not where retailers expect their payback.

In Brief

• Anthropic accuses Alibaba of extracting Claude's capabilities Anthropic said Alibaba illicitly pulled model capabilities from its Claude systems. The accusation names a major Chinese cloud rival and raises the prospect of legal or export-policy fallout.

• Amazon commits $13B to AI data centers in India Amazon pledged a fresh $13 billion for AI infrastructure in India. The spend lands as US cloud providers race to build compute capacity in the country.

• Paying AI users are picking Claude over ChatGPT Spending data shows consumers who pay for AI increasingly choose Anthropic's Claude. ChatGPT still holds the larger overall user base, but Anthropic is gaining share among subscribers.

• Adobe buys Topaz Labs for image and video upscaling Adobe acquired Topaz Labs, maker of AI enhancement and upscaling tools. Adobe plans to fold the technology into its existing apps.

• Agility Robotics plans a $2.5B SPAC listing Agility Robotics, the Oregon State humanoid spinout, will go public through a SPAC at a $2.5 billion valuation. The deal is expected to generate $620 million in proceeds.

• General Intuition raises $320M to train agents on gameplay General Intuition closed $320 million to train AI agents on millions of hours of video game footage. The company bets action data from games transfers to real-world tasks.

• Europe resists US push to widen China chip curbs European officials and ASML are pushing back on Washington's proposed export limits. The MATCH Act would block sales of older deep-ultraviolet tools that China can already buy.

• Databricks' former AI chief claims a 1,000x cut in AI power use Databricks' ex-AI lead launched a startup promising to slash AI energy costs by three orders of magnitude. Its first system, Un-0, generates images and aims to replicate conventional AI workloads at lower power.

• Patronus AI raises $50M to stress-test AI agents Patronus AI, founded by former Meta researchers, raised $50 million to build simulated environments that probe agents for failures. Its investor cited heavy enterprise demand for agent testing.

• Notion shuts its email app to push inbox agents Notion is killing the Skiff-derived email app it acquired. The company said most users now rely on AI agents to handle their inbox instead.

• Former Infosys CEO launches an IT services challenger Vishal Sikka started a new venture to compete with traditional IT services firms. Mayfield and Aramco Ventures backed the company, which recruited veterans from SAP, Infosys, and VianAI.

• iLLaDA trains an 8B diffusion language model from scratch Researchers built iLLaDA, an 8B masked diffusion model with fully bidirectional attention, on 12 trillion pretraining tokens. The work tests diffusion as an alternative to autoregressive training at production scale.

Read the full edition →