Apple Siri picks Google Gemini; Detroit cuts 20,000 in AI pivot

Per a TechCrunch report citing leaked product specs, Apple will unveil at WWDC in June 2026 a rebuilt standalone Siri app whose chatbot experience runs on Google Gemini — not Anthropic Claude or OpenAI ChatGPT, despite reporting earlier in May that both labs were courting Apple as a launch partner. The app's headline feature is auto-deleting conversations with user-configurable retention windows of 30 days, one year, or indefinite, mirroring the privacy controls in Apple's Messages app. Analysts read the privacy emphasis as marketing cover for Apple's continued backend dependence on a competitor's frontier model, three years after the original Siri-on-ChatGPT integration shipped at WWDC 2024.

Citing CNBC calculations, this week's TechCrunch Mobility newsletter reports that Ford, General Motors, and Stellantis have together eliminated more than 20,000 U.S. salaried positions — 19% of their combined office workforces — since the AI restructuring wave began. GM specifically eliminated 600 IT roles in the past week to make room for hires in "AI-native development, data engineering, cloud engineering, agent and model development, and prompt engineering." The cuts mirror Cisco's roughly 4,000 layoffs covered in this newsletter on May 14 — legacy enterprises are systematically pivoting headcount from traditional IT to AI engineering, and three of America's largest manufacturers are now doing so in tandem.

Stratechery's Sunday essay tackles the political backlash forming around hyperscale data-center sites — a theme that's recurred across this week's newsletter (NAACP versus xAI in Mississippi, the Pennsylvania town hall, Peter Thiel's bet on offshore wave-powered data centers via Panthalassa). Ben Thompson's framework: local opposition is rational, because the upside (compute capacity, AI capability) accrues to coastal labs and shareholders while the downsides (noise, water draws, transmission lines, falling property values) hit the communities directly. His proposed fix is the politically uncomfortable but mathematically simple one — pay opponents directly rather than promising jobs that don't materialize.

A Show HN post on Sunday morning highlighted Semble, an open-source code-search library by MinishLab purpose-built for AI coding agents. The pitch: while grep-plus-read approaches need a full 100,000-token context window to hit 85% recall on benchmark queries, Semble hits 94% recall at just 2,000 tokens — roughly a 50-times reduction in tokens billed per query. The architecture combines tree-sitter-based code chunking, Model2Vec embeddings (potion-code-16M), BM25 lexical retrieval, and reciprocal-rank fusion. Indexing a typical repository takes around 250 milliseconds; query latency clocks in at 1.5 milliseconds, with the latest v0.1.7 release shipping on May 12.