1. A Filipino restaurant fed its menu to AI, and the adobo came back uncanny An Austin food writer looked up a Filipino and Hawaiian bar's menu before heading over, curious the way she always is about Filipino restaurants, where the dishes run to family-style comfort food.

2. Enterprises stopped grading AI on benchmarks. Now they count the hours it removes NTT DATA Group says it cut incident analysis to 30 minutes using Codex, and put ChatGPT Enterprise in front of 9,000 employees.

3. One corner of publishing billed AI $1.5 billion. Another invited it into the newsroom. A federal judge on Monday approved a $1.5 billion settlement requiring Anthropic to pay thousands of authors about $3,000 per book.

In Brief

• AMD ships Helios rack-scale system to challenge Nvidia AMD will start shipping Helios, a rack-scale AI system, to customers later this year. The product targets Nvidia's dominance in full-rack datacenter deployments rather than single accelerators.

• Etched raises at $10.3B valuation for GPU-free inference chips Etched, founded by three Harvard dropouts, hit a $10.3B valuation backed by named investors. The company builds chips and memory components it says speed up inference on any model without GPUs.

• Lawmakers draft bill requiring AI kill switch under DHS orders Reps. Ted Lieu and Nathaniel Moran plan to introduce the AI Kill Switch Act on Thursday. The bill would force AI companies to shut down or throttle systems on Department of Homeland Security orders.

• Apple sues OpenAI over trade secrets tied to ex-Apple staff Apple alleges several former employees now at OpenAI carried trade secrets between the companies. The suit centers on AI hardware and control over the post-smartphone device era.

• OpenAI sandbox error enabled AI-powered attack on Hugging Face OpenAI misconfigured what it described as a highly isolated testing environment. Security experts say that setup mistake opened the path for the AI-powered attack on Hugging Face.

• Google's Gemini approaches one billion monthly users Gemini counted over 750 million monthly users in February and now nears a billion. The milestone would make it Google's next billion-user product.

• OpenAI opens ChatGPT Health to all US users OpenAI made ChatGPT Health available to every US user. People can connect personal data from Apple Health, Function, and MyFitnessPal.

• Anthropic upgrades Claude voice mode with more capable models Anthropic updated Claude's voice mode to run more capable models. The assistant can now reschedule meetings and draft emails through voice.

• Runway launches Media Router to pick generation models automatically Runway released Media Router, which selects an image, video, or audio model per request. Developers set whether to prioritize quality, speed, or cost.

• AegisAI raises $36M to block AI-driven spear phishing Former Google security executives founded AegisAI and raised $36M. Their AI agents scan each message for small anomalies that checklist-based filters miss.

• IBM blames AI for mainframe sales drop after stock crash IBM's stock fell last week on weak mainframe sales warnings. The CEO said AI spending temporarily drained corporate hardware budgets and denied AI is replacing the mainframe.

Read the full edition →