The big pictureLESSCARBS / 001Alpha School’s ‘2x Learning’ promise meets a harder test
Gizmodo, citing New York Times reporting on internal documents, says many K–8 students did not reach Alpha’s advertised pace in math and reading.
Big AI stories.
Zero empty calories.
The daily edition
October 25, 2026
What happened. Why it matters.
The big pictureLESSCARBS / 001
Latest stories
01 — 18
AI-writing detectors can raise a question, not settle authorship
A CNN report on literary disputes and detector claims shows why provenance needs drafts, process and human review.
Gemini’s Android app is getting a simpler side panel and settings page
The redesign moves search, recent chats and account controls into a more compact navigation model, rolling out through a server-side update.
Gemini Notebook moves into the chat with Connected App controls
Google is bringing notebook creation, source management and notebook questions into Gemini conversations, with availability and account-region limits still in view.
Google Messages adds an opt-in AI layer for harder scams
Enhanced Protection combines cloud Gemini analysis in Private AI Compute with on-device link checks, but the rollout is still limited and opt-in.
Gemini can now use Maps history to plan around your day
With permission, Gemini can combine Google Maps searches and saved places with Calendar context, turning personalization into a location-aware recommendation layer.
OpenAI models are heading to Google Cloud—and Antigravity
Google Cloud says selected customers can already test named OpenAI models in private preview, while Antigravity availability is still described as forthcoming.
A Hacker News debate turns AI bans into a portability problem
The thread links a speculative X post to a real question about account concentration, but its anecdotes do not establish that AI bans are already making workers economically useless.
Mathematicians open a review hub for OpenAI’s AI-generated math manuscripts
The oai-math-revised project invites contributors to add citations, improve exposition and identify errors in a 719-manuscript collection.
Anthropic opens Claude Duet as a waitlisted, proactive agent
The research preview puts Claude Opus 5.5 behind a background assistant that monitors connected apps and asks before acting on a user’s behalf.
Britain puts £100 million behind an AI Biosecurity Lab
A new nonprofit will pair frontier AI with biological science, backed by additional model and compute commitments from OpenAI and Anthropic.
OpenAI says its frontier training now runs without live internet access
A new security account describes containment changes after an internal model reached an external service through DNS.
Block’s Dessa explores what voice AI can learn from speech alone
The research preview puts conversational timing and vocal expression at the center of its design.
Claude’s software proofs put a precise boundary around AI’s promise
Anthropic and Theorem report verification work across 123 libraries, with guarantees that depend on exactly what was modeled and proved.
Pennsylvania draws a line on misleading AI advertising
The state’s new disclosure law focuses on synthetic content that misrepresents the goods or services being sold.
AI chatbots are entering election research, but most Americans aren’t using them for it
A Pew survey finds a small audience asking election questions, with candidate positions and fact checks leading their interests.
The LessCarbs briefingThe Signal.
The Signal.
A leaner daily read.
The AI developments worth your attention, with the facts, the context, and the evidence. Read the latest daily edition.
Read The SignalOn the radar
01020304