
OpenAI eyes 2028, Meta buys Conductor, Google vs. schools
OpenAI starts playing political chess, Meta snaps up an agentic coding startup, and Google picks a fight with every school district in America.
Subscribe free →
OpenAI Is Already Lobbying for the 2028 Election — and Its Own Bottom Line
OpenAI policy chief Chris Lehane is pitching a "new social contract" around AI, explicitly framing it around the 2028 election cycle. The proposal promises economic benefits for Americans but conveniently maps onto outcomes that keep OpenAI at the center of AI policy.
Why it matters: A major AI lab openly trying to shape electoral politics two years out is a new frontier — and a preview of how much harder AI governance debates are about to get.

Meta Buys Conductor to Field Entire Agent Teams for Coding
Meta has acquired Conductor, an AI coding startup that lets engineers spin up multiple specialized agents to tackle different parts of a project in parallel — not just one assistant, but a whole coordinated crew. Terms weren't disclosed.
Why it matters: This signals Meta is betting hard on multi-agent workflows as the future of software development, not just copilot-style autocomplete.

Google Threatens to Kill Search for Schools That Block Its AI — and Everyone Hates It
Google announced it would disable Search access for schools that choose to restrict its AI features, triggering immediate backlash from NYC mayor Zohran Mamdani, former FTC chair Lina Khan, and others who called it a blatant abuse of monopoly power.
Why it matters: Using search access as a lever to force AI adoption on schools is exactly the kind of bundling behavior that's already under antitrust scrutiny — this could accelerate regulatory action against Google.

Anthropic Wants to Give AI Models a Verifiable Identity
Anthropic is developing a system called FlashProve that lets anyone cryptographically verify whether an AI response actually came from the model it claims to be from — a kind of digital fingerprint for AI outputs.
Why it matters: As AI-generated content floods the web and impersonation attacks grow, being able to prove an output came from a specific, known model is a foundational trust primitive the industry has been missing.

Data Centers Are Strapping Jet Engines to the Grid Because They Have No Choice
The years-long wait for utility grid connections and new turbines has gotten so bad that data center operators are resorting to repurposed jet engines — less efficient, dirtier, and more expensive than conventional power, but available now.
Why it matters: The AI compute buildout is so far outpacing energy infrastructure that operators are making environmentally ugly stopgap decisions — a problem that compounds the longer grid permitting stays broken.

ChatGPT Can Now Generate Animated GIFs — Powered by Images 2.6
OpenAI rolled out native GIF generation in ChatGPT via its new Images 2.6 model. You can generate them from a text prompt or uploaded photo, tweak animation speed, edit individual frames, and share directly.
Why it matters: It's a small feature on its surface, but it shows how quickly OpenAI is closing the gap between ChatGPT and specialized creative tools — and raises the floor for what people expect from a general-purpose AI assistant.
Quick Hits
- →DuckDuckGo is capitalizing on the Google-schools blowup by courting frustrated educators with its AI-optional search engine. TechCrunch
- →Apple's new Siri Recaps feature got a surprisingly warm review from a skeptical TechCrunch writer — useful enough to win over a privacy hawk, apparently. TechCrunch
- →Nothing CEO Carl Pei teased a screenless AI device launching in December, described as "a tool, but one built for humans." The Verge
- →AT&T's Connected Life home security system now integrates with Gemini for Home, letting the AI summarize events and search your video history. The Verge
- →Waymo is ending its Uber partnership in January 2028, setting up direct head-to-head competition in overlapping markets. TechCrunch
AI TLDR