1. Google and Anthropic Ship Near-Identical "Reusable Prompt" Features Days Apart Google's Chrome browser now lets users save any Gemini prompt as a "Skill": a reusable command that runs across any open tab with one click.

2. The People Building AI Think It's Safe. The Public, Stanford Found, Disagrees on Nearly Every Measure. The people building AI systems and the people affected by them cannot agree on basic facts about the technology.

3. One Developer Claims He Can Strip and Forge Google's SynthID Watermarks A software developer going by the handle Aloshdenny published a GitHub repository last week. It contains code, documentation, and a claim that tests one of Google's most promoted AI safety tools.

In Brief

• OpenAI Acquires Personal Finance Startup Hiro OpenAI bought Hiro, an AI-powered financial planning company. The deal signals OpenAI's intent to embed financial planning directly into ChatGPT. TechCrunch

• Google DeepMind Releases Gemini Robotics-ER 1.6 for Autonomous Robot Control DeepMind shipped Gemini Robotics-ER 1.6, upgrading spatial reasoning and multi-view scene understanding for physical robots. The model targets autonomous task execution without per-task fine-tuning. DeepMind

• Science Corp. Prepares First Human Brain Sensor Implant Max Hodak's Science Corp. plans to place its first sensor in a human brain. Early applications include electrical stimulation of damaged brain and spinal cord cells to promote healing. TechCrunch

• UK Government's Mythos AI Completes Multistep Cybersecurity Infiltration Challenge A UK government AI model called Mythos became the first AI system to finish a difficult multistep cyber infiltration test. The benchmark aims to measure actual offensive AI capability against real-world attack scenarios. Ars Technica

• Ukraine Deploys Ground Robots to Replace Soldiers in Drone Kill Zones Ukraine is scaling military robot deployments to reduce human exposure in areas saturated by drone strikes. The shift moves ground combat toward mixed human-robot units. Ars Technica

• Unitree Lists $4,370 Humanoid Robot on AliExpress Unitree is selling its R1 humanoid robot internationally through AliExpress at $4,370. The robot ships with acrobatic movement capabilities, though practical consumer use cases remain undefined. Wired

• Anthropic's Long-Term Benefit Trust Adds Novartis CEO Vas Narasimhan Anthropic appointed Vas Narasimhan, CEO of Novartis, to the board of its Long-Term Benefit Trust. The trust oversees Anthropic's public benefit mission independently from the company's commercial leadership. Anthropic

• US Hospitals Roll Out AI Chatbots in Patient Portals Hospital systems across the US are deploying AI chatbots inside patient portals to handle health questions. The expansion comes as more Americans already use consumer AI tools for medical advice. Ars Technica

• Vibe-Coding App Anything Pivots to Desktop After Two App Store Removals Anything, an app that lets users build mobile apps through natural-language prompts, is launching a desktop companion after Apple removed it from the App Store twice. TechCrunch

• Silicon Valley Spends Millions to Block Former Palantir Engineer's Congressional Bid Alex Bores, a former Palantir employee turned New York state legislator, helped pass one of the country's strictest AI laws. Major tech companies are now funding opposition to his congressional campaign. Wired

Read the full edition →