Google will switch off Search in schools that turn off its AI

Google will switch off Search in schools that turn off its AI

Google announced on Tuesday that it plans to disable AI tools for K-12 students next year unless school administrators explicitly turn them on. The change covers Gemini, NotebookLM, Gemini in Classroom and Google Search¹. Schools that restrict AI altogether will lose access to Google Search as well, because AI features are now part of the search engine^1,2.

Administrators will not get an option to switch off the Gemini-powered features while keeping Search². Chris Phillips, Google's vice president for education, told The New York Times that Search and AI are deeply integrated. Students in schools that block AI will see a message from their school saying Search is not available^1,2. Google says it will share more specifics with schools in the coming months, before the changes roll out¹.

Teachers may not want students to use AI Overviews for research or assignments, but they likely still want them to reach the search engine for sources. TechCrunch says the decision shows Google's influence over schools, given how common its apps and Chromebooks are in classrooms.¹.

“Search and AI are deeply integrated”

Apple opens Siri AI to third-party apps as it hunts for new revenue

Apple will let Siri AI perform actions across third-party apps. As part of the iOS 27.3 update rolling out on Tuesday, developers can now build Siri AI's new features into all apps. The update is rolling out as part of a Developer Preview⁴. The day before, on Apple's earnings call, its finance chief pointed to third-party apps as a possible source of AI revenue³.

Siri AI can draw on a user's personal context, understand natural language requests and use on-screen awareness to complete tasks. A new app skills feature lets developers combine several tasks into a single action.⁴. Chief financial officer Kevan Parekh said there are opportunities across third-party apps and partnerships, but did not say what they are³.

Apple's App Store has faced regulatory pressure and a slump in mobile gaming, which have hurt services growth. Third-party estimates in August suggested an 18% drop in Apple's U.S. commission revenue.. Apple expects services growth of 8 to 9% in the December quarter³.

Anthropic's revenue jumps, but compute now eats 58% of it

Anthropic released new financial data in a filing on Monday, an update to the S-1 it filed last week.. Revenue reached $17 billion in the quarter, up from $11.5 billion in the quarter before. Its operating loss also grew, largely because of rising compute costs⁵.

Compute took up 58% of Anthropic's revenue in the new quarter, up from 48% the quarter before. Revenue for every dollar spent on compute fell to 1.72x, after peaking at just over 2x. The compute bill reached $10 billion for the quarter. That included $5.0 billion of compute under short-term commitments, which the filing says costs significantly more than compute bought under multi-year deals⁵.

TechCrunch says demand grew so fast that Anthropic had to scramble for compute, which meant paying higher rates. It has been paying SpaceX $1.25 billion a month for compute, alongside multibillion-dollar deals with Nscale, Akamai and Volta.. The filing says it expects compute and infrastructure costs to keep increasing in absolute dollars⁵.

Anthropic is working on a way to prove which model wrote an answer

Anthropic is working on an open source system to verify which AI model produced a given output, according to a blog post published Tuesday. The system is called FlashProve. It is built to provide a cryptographic proof that a given output came from a specific model. The aim is to let users confirm that an answer really came from the model they think they are using⁶.

The system requires AI labs to publish a fingerprint of each model they release, which can later be matched to specific outputs. The fingerprints don't reveal a model's weights, the internal numbers that define how it behaves. That means closed-model makers could publish them without giving up their most valuable asset. The system is designed to be lightweight, so checks could run on a local device⁶.

The researchers say it could stop an AI company from secretly swapping in a cheaper model to cut costs.. Research is ongoing, and there are no plans to deploy it at Anthropic or any other lab⁶.

A Postman user says it installed plugins into their coding agents

A user opened a GitHub issue saying the Postman desktop app automatically installed plugins into their coding agents, Codex and Claude Code. The user, indicava, says the changes followed opening Postman on a Mac. They describe it as a severe consent and trust problem.⁷.

The user says they found a user-wide Postman plugin, a startup instruction hook, a remote MCP connection, a Claude Code plugin and a separately installed Postman CLI. They quote Postman's own log, which records claude-code and Codex as installed by a script.. It also says the installer tried to mark its own startup hook as trusted in the Codex settings file⁷.

According to the report, the hook tells the agent to route API engineering tasks through Postman. The user argues that an API client should not make itself the default workflow in other tools simply because it was opened.⁷.