Google schools: AI off by default — and Search can't be had without it

Share
Google schools: AI off by default — and Search can't be had without it

Three stories about the same thing from different angles: who controls the defaults, who pays for the capacity, and who is now lending the money. Start with the defaults, because Google just moved the most consequential one in American classrooms.

Google will switch its AI products off by default for K-12 student accounts, and the one product schools cannot replace — Search — comes with its AI attached. In an education announcement on Tuesday, Chris Phillips, Google's vice president and general manager for education, wrote that access to Gemini, Gemini Notebook, Gemini in Classroom and Google Search "will be entirely off for K-12 student accounts until school administrators explicitly turn them on," per Google's announcement. That is a genuine reversal: through the summer, Google's education suite defaulted students into AI features, and districts that objected had to opt out product by product. Now the burden flips to the school.

The coverage mostly reported that Google will cut off Search for schools that reject AI, which is true and is not the whole mechanism. Search and its AI cannot be separated. Phillips told the New York Times that "Search and AI are deeply integrated," so a school that blocks AI entirely will not have Search for its students, who instead see a message from the school saying it is unavailable. But the other three products toggle independently, and Google told Engadget the arrangement is "not an all-or-nothing deal." A district can keep Search, and with it AI Overviews and AI Mode, while disabling Gemini, Gemini Notebook and Gemini in Classroom for its students. What no district can have is a search engine with the AI surgically removed.

The AI that replaces the old experience is narrower than the consumer product. For students signed in with school accounts, Guided Learning replaces AI Overviews, and inside Google Classroom Guided Learning is the only version of Gemini available — students cannot switch back to the regular model, and teachers can review chat transcripts, Chalkbeat reported. Guided Learning in Classroom is live globally now. The Search version starts as a pilot in a small number of U.S. schools this month, with the wider control changes due in the coming months; Chalkbeat and Quartz put the broad rollout early next year, while Google's own post commits to no date beyond telling schools more "ahead of the changes rolling out."

Scale is the reason this matters more than a settings-page edit. Google Classroom is used by more than 150 million teachers and students, by Google's own count, and in a Chromebook district the alternative to Google Search is not Bing, it is a procurement project. That asymmetry is what critics seized on. Lina Khan, the former Federal Trade Commission chair, told the Times that Google is using its "search monopoly to coerce schools," and Common Sense Media's Eisha Bush told Chalkbeat the likelihood of schools and students avoiding Google Search is "slim to none" — her organisation rates AI Overviews and AI Mode an "unacceptable risk" for students.

Read the sequence and the strategy is legible. New York City barred student-facing AI through eighth grade in September, Los Angeles did something similar, and the backlash to default-on AI in classrooms has been building since the spring; we covered the New York decision at the time — New York City bans student-facing AI through eighth grade. Google's answer is not to defend the old default but to hand schools a switch and price the switch honestly: AI-on with Search, or AI-off without it. Defaults are the product in consumer software, and Google just made the default the argument. The districts that want out will discover the exit is real and narrower than the press release implied.


Fal raised a $1.3 billion Series F led by Bessemer Venture Partners at a $16 billion valuation, and disclosed a $350 million Series E it closed in April at $8 billion — $1.65 billion combined, doubling its mark in seven months. The company announced both rounds on Tuesday, per fal's own post, with NVIDIA joining as a strategic investor alongside Sound Ventures, HOF Capital and T. Rowe Price; Bessemer's Talia Goldberg takes a board seat. Bloomberg, which broke the numbers, reported that the Series F closed in October. Fal confirmed the combined total and the leads but published neither the valuation nor a per-round split, so the $16 billion figure rests on Bloomberg's reporting.

The revenue line is the part to handle carefully. Bloomberg attributes $800 million in annualised revenue to a person familiar with the matter, and The Information reported the same figure in September, up from roughly $400 million in March. Fal has not confirmed it. If it holds, the round prices the business at about 20 times revenue — against Baseten raising at up to $30 billion, Fireworks AI in talks at $30 billion to $40 billion and Modal at $15.75 billion, all reported this autumn. Fal is cheaper than the LLM-infrastructure marks, and there is a reason: it sells inference for generative media, not text.

That business is bigger than it sounds. Fal runs a single API over more than 1,500 image, video, audio and 3D models — Black Forest Labs' FLUX line, Google's Veo, ByteDance's Seedance, Kling, MiniMax — plus post-trained exclusives built with the labs, dedicated GPU clusters and serverless deployments where customers keep their own weights. It claims 2.5 million developers and names Adobe, Canva, Shopify, Amazon and World Labs as production customers.

Chief executive Burkay Gur gave Bloomberg the sentence that explains the raise better than any multiple: "We hit a point in summer where we were not able to serve our customers and had to rate limit them." In media inference, demand arrives in spikes tied to whichever model went viral that week, and the only cure is owned or reserved GPUs. This round is a capacity bet, not a research bet — which also means the risk sits upstream. Fal resells access to models owned by Google, ByteDance, Black Forest Labs and MiniMax, every one of which could meter or disintermediate a reseller. At 20 times an unconfirmed revenue number, investors are paying for the queue, not the catalogue.


Meta is selling between $20 billion and $25 billion of bonds, two-thirds in dollars and one-third in euros, its first euro-denominated offering and its second large debt sale in six months. Pricing was expected Tuesday afternoon, CNBC reported, citing people familiar with the matter; Bloomberg first reported the plans, and Meta declined to comment.

The balance sheet is why this is an AI story rather than a treasury note. Meta guided full-year capital spending to $135 billion to $145 billion, posted negative free cash flow of $2.12 billion in the third quarter, and said last week it plans to raise $95 billion in total — $35 billion through underwritten equity offerings and $60 billion through an at-the-market programme. On Friday, S&P Global Ratings cut its outlook on Meta's AA- rating to negative, citing uncertainty over the capex plan and the negative cash flow. Meta's stock rose 2.7% on Tuesday after closing out its worst month in four years.

Six months ago the same company raised $25 billion in bonds without anyone calling it a warning. The difference now is that the debt is funding a buildout whose returns are not yet visible in free cash flow, and the rating agency has said so in writing. When a company with Meta's cash flow starts terming out AI capex in euros, the marginal dollar of compute is being priced by creditors rather than by shareholders — and creditors do not care how good the demo was.

What to watch: whether districts publish opt-out lists, and whether the Search pilot's first schools say yes. On the money side, watch Fal's GPU commitments and Meta's December capex update — the two ends of the same trade.

Would your local school trade Google Search for an AI-free classroom? Write to us via our contact page.

Get SiliconNoon in your inbox

One email a day with the AI stories that matter — what happened, the numbers, and why it matters. No spam, unsubscribe in one click.

Read more

Treasury's outbound investment rule adds supercomputing, hypersonics

Treasury's outbound investment rule adds supercomputing, hypersonics

The US Treasury widened the list of technologies American money is not supposed to fund abroad, and asked the market what it thinks of the list. The question underneath is whether screening private capital does anything that export controls on hardware did not. The Treasury Department proposed a rule on Thursday that extends its outbound investment screening to high-performance computing, supercomputing, hypersonic systems and biotechnology, on top of the semiconductors, quantum and…

Senate report: hyperscalers misled the public on data center costs

Senate report: hyperscalers misled the public on data center costs

A yearlong Senate investigation just put Congress's own numbers under the claims AI data centers sell to towns and ratepayers. Also: the Times finally counts what Anthropic's agents submitted to the State Department, and StepFun sets an open-weights date for its top-ranked flagship. A yearlong Senate investigation led by Senators Warren, Van Hollen, and Blumenthal concludes that some of the biggest hyperscalers misled the public about what their AI data centers cost everyone else.

A judge calls Flock's plate searches mass surveillance

A judge calls Flock's plate searches mass surveillance

A landmark Fourth Amendment ruling lands on the same day Congress moves to cut license-plate readers out of federal hands — and in the funding lane, the database layer agents build on just got its biggest raise yet. A federal judge ruled that a warrantless search of Flock's license-plate network was an unconstitutional "type of indiscriminate mass surveillance." U.S. District Judge Sara Hill of Northern Oklahoma suppressed the evidence in United States v.

OpenAI backed a startup built to fight an AI-made pandemic

OpenAI backed a startup built to fight an AI-made pandemic

A biosecurity startup raised $36 million to design antibodies against the pathogens AI might help build — and it was already quietly firing at the attack surface it is defending. Separately, DHS started using AI to recommend what to redact from the records you asked for, and an anonymous model took third place on the world's busiest token leaderboard with a fingerprint no one was supposed to see. Red Queen Bio, a biosecurity startup whose $15 million seed OpenAI led in November, has now raised…

Newsom orders a kill switch and onsite auditors for frontier AI

Newsom orders a kill switch and onsite auditors for frontier AI

California Gov. Gavin Newsom signed an executive order Friday that tells his own agencies to design the state's next round of AI rules — and names the two things the industry has resisted hardest: a kill switch for frontier models, and independent auditors embedded inside the labs. Executive Order N-9-26 gives the Government Operations Agency until November 16, 2026, to deliver recommendations on four specific mandates, and each one aims at the same target.

Nadella backs AI pacing, and picks a fight over who referees

Nadella backs AI pacing, and picks a fight over who referees

The pacing camp got its fourth member tonight, and the newest one arrived with an objection attached: Microsoft's CEO agrees the frontier should slow down, and does not agree that Anthropic, OpenAI and Google should be the ones deciding what that means. Satya Nadella wrote on X shortly after 7:30 PM ET that Microsoft "welcome[s] the research, focus, and deliberate pacing needed to get alignment right as the design goal," and announced a "Code of Conduct" governing Microsoft's own first-party…