AI News Today (Nov 5, 2026): Data Centers, Google Schools & More

AI news today, November 5, 2026: data centers lost the midterms, Google is fighting a school Search backlash, and OpenAI agreed to buy Pierre Computer. Here are the 10 top stories and 5 security stories, each with free, non-paywalled sources.

Key Takeaways

• Data centers lost the midterms: 33 of 45 local ballot measures will restrict new development this cycle.
• Google says schools can reject AI but then lose Search, drawing criticism from Bernie Sanders and Zohran Mamdani.
• OpenAI will acquire Pierre Computer and fold its Code Storage service into the Codex coding platform.
• AMD data center revenue hit $8.5 billion, up 97%, as total quarterly revenue reached $13.4 billion.
• WorkOS says the ARTEX AI framework breached an internal sales app, exposing CRM and contact data.

1. Update: Google Schools Face Backlash Over Search-AI Opt-Out

Google is responding to political backlash after reports that it would disable Search for schools that reject its AI tools. The company said Gemini, Gemini Notebook and Search are off by default, and schools can turn one or all of them back on. We first covered the policy in our November 4 briefing.

Critics including Bernie Sanders, Zohran Mamdani and former FTC chair Lina Khan called the policy extortion and monopoly abuse. Google education vice president Chris Phillips told The New York Times that Search and AI are "deeply integrated." Google says AI Overviews will add guided-learning steps.

Why it matters: Schools now face a package deal: accept classroom AI or lose the search tool students and teachers already use.

• Source: Google responds to political backlash over removing search from schools that ban AI — CNBC
• Source: Google faces backlash for plan that would turn off Search for schools that restrict AI — TechCrunch

2. Update: Data Centers Lose the Midterm Elections

Voters rejected data center expansion across the country. In 45 local ballot measures, 33 will restrict development, according to Redbud Consulting data cited by The Verge. NBC's Election Day poll found 76% of voters opposed data centers in their communities. We first covered the backlash in our October 24 briefing.

Texas elected Gina Hinojosa, who backs a statewide moratorium. Michigan elected Abdul El-Sayed, who wants a pause on new projects. Ohio returned Sherrod Brown, who attacked his opponent over data centers. Some pro-development candidates still won, including Florida Governor-elect Byron Donalds. Wells, Maine, backed a 180-day moratorium on large data centers.

Why it matters: AI's power bottleneck is becoming a political bottleneck, and developers now face local votes before they face grid queues.

• Source: Data centers lost the midterms — The Verge
• Source: The Results Are In: Voters Hate Data Centers — Wired

3. OpenAI to Acquire Pierre Computer for Codex Tools

OpenAI will acquire Pierre Computer Company, the startup behind Code Storage and the open-source Trees, Diffs and Highlights libraries. Trees and Diffs already power file browsing and change review in the Codex app. Code Storage lets apps store code and track its history with Git. Financial terms were not disclosed.

OpenAI plans to fold Code Storage into Codex after closing and keep Pierre's open-source projects open. Existing Code Storage customers can keep using the service during the transition. The deal needs regulatory approval, and both companies will remain separate until then. It follows OpenAI's Pando source-control project, disclosed earlier this week.

Why it matters: OpenAI is buying the plumbing that stores and reviews code written by agents, not just the models that write it.

• Source: OpenAI to acquire Pierre Computer Company, Inc. — OpenAI
• Source: OpenAI is acquiring software startup Pierre Computer Company — The Verge

4. AMD Data Center Revenue Jumps 97% to $8.5 Billion

AMD reported third-quarter revenue of $13.38 billion, up 45%, and data center revenue of $8.5 billion, up 97%. Adjusted earnings per share were $2.09. Data center now supplies 64% of AMD's revenue, up from 47% a year earlier. Operating income reached $2.8 billion as data center sales grew.

AMD guided fourth-quarter revenue to about $17 billion, above the $16.2 billion analyst average. GAAP net income was $2.4 billion, or $1.46 per share. CEO Lisa Su said server revenue should double in 2027 as Helios racks ramp. The stock rose 8% after hours before paring gains as election results arrived.

Why it matters: AMD is turning AI demand into a second GPU franchise, giving hyperscalers a real alternative to Nvidia.

• Source: AMD Reports Third Quarter 2026 Financial Results — AMD
• Source: AMD's data center revenue nearly doubles in Q3 and CEO Su expects growth to accelerate — CNBC

5. Canva Cuts or Relocates 1,271 Roles as Growth Slows

Canva will cut, change or relocate 1,271 roles, about 22% of its workforce, CEO Melanie Perkins said. The company's growth slowed for the first time in a decade, and it lowered its 2026 revenue growth forecast from 30% to 20% in August. Perkins called it the hardest moment in Canva's 13-year history.

Affected staff get at least six months' salary, 12 months of healthcare where offered, and $10,000 grants to start businesses. Canva will merge its product, growth and platform teams and make Sydney the primary home for more product roles. The changes take effect from January.

Why it matters: AI design tools are compressing Canva's growth just as it tries to justify a long-delayed public listing.

• Source: Canva layoffs: design giant to cut or relocate more than 1,200 jobs in major restructure — The Guardian
• Source: Canva CEO says a major reorganization will affect 22% of its workforce. Read the memo. — Business Insider

6. Big Tech's Hidden AI Debt Swells to $3.2 Trillion

Off-balance-sheet AI commitments at Alphabet, Amazon, Meta, Microsoft and Oracle reached about $3.2 trillion at the end of September, Nikkei reported. That is up 90% in six months and equals roughly 80% of the five companies' combined assets. Nikkei said some figures were estimates and methodology changed.

Long-term purchase agreements for chips, servers and power make up 60% of the total. Leases that have not started make up 40%. Alphabet carries $1.05 trillion and Meta $810 billion. Goldman Sachs separately estimated a similar $3.2 trillion total. The commitments could translate into large cash outflows as contracts begin.

Why it matters: The AI buildout now depends on future obligations that stay off today's balance sheet, so investors must price a risk they cannot see.

• Source: U.S. Big Tech's AI Off-Balance-Sheet Commitments Surge 90% to $3.2 Trillion — Businesskorea
• Source: Big Tech has more than $3 trillion in AI commitments that don't show up on its balance sheets — Yahoo Finance

7. UK's £400 Billion AI Figure Came From Google Report

The UK government's £400 billion AI boost estimate came from a 2023 report commissioned by Google and produced by Public First, AI minister Kanishka Narayan said in a parliamentary answer. The government's own central estimate is £120 billion by 2030, less than a third of the headline figure.

Narayan said AI adoption and data investment could add £55 billion to £140 billion, a "hugely uncertain" range. Departments used the £400 billion figure for nearly two years, and Skills England was still publishing it four months after ministers received the lower estimate. The answer did not withdraw the larger figure or call it wrong.

Why it matters: AI forecasts are becoming policy commitments, and the UK case shows how a vendor-funded number can outrun the government's own evidence.

• Source: UK Officials Said AI Would Add £400 Billion to the Economy. That Figure Mainly Came From Google — Wired
• Source: The £400 Billion Figure Was Never the Government's Own — freegardner

8. Nvidia Lets AI Agents Steer SETI's 42 Dishes

The SETI Institute and Breakthrough Listen are using AI agents to steer the Allen Telescope Array in Northern California. A single natural-language request can point the 42 dishes, correct incoming data and run analysis models on the live stream. Researchers can also monitor packet loss and rewrite signal-processing blocks by prompting.

Nvidia's Holoscan platform handles up to 2.4 terabits per second on RTX PRO 6000 Blackwell GPUs. The system made the first real-time AI detection of a pulsar in the Crab Nebula. Researchers still type the commands, and no alien signal has been found. Next, the team is evaluating open models for offline observatories.

Why it matters: AI is moving from analyzing telescope data after the fact to operating the instrument while it observes, a model that could extend to lunar observatories.

• Source: Things That Go Beep in the Night: Agentic AI Joins the Search for Extraterrestrial Signals — Nvidia
• Source: AI agents take over at SETI's Allen Telescope Array, scanning for extraterrestrial intelligence — The Decoder
• Source: The Allen Telescope Array (ATA) — SETI Institute

9. Uber Fulfilled Just 4 Rides From AI Agents in Q3

Uber's rides business fulfilled four trips booked by third-party AI agents in the third quarter, CFO Balaji Krishnamurthy said on the earnings call. Uber Eats handled about 500 agent orders during Halloween week, out of roughly 320 million weekly trips and deliveries. Uber reported the figure on its Q3 earnings call on November 4.

Krishnamurthy called agent demand "very, very early and very limited." It was Uber's first disclosure of agent-originated volume, and it landed as platforms weigh whether assistants will disintermediate their apps. Uber has not said which agents or protocols generated the demand.

Why it matters: The agentic-commerce boom is still mostly a demo, and the first hard number from a major marketplace is a rounding error.

• Source: Uber only fulfilled 4 rides from AI agents last quarter — The Verge
• Source: Uber Fulfilled Just 4 Rides From AI Agents — Tech Insider

10. Hinton Says Humans May Have Been Better Off Without AI

Geoffrey Hinton told The Guardian that "we would have been better off never developing" AI. The Nobel laureate warned of a Chernobyl-type incident this year or next, such as a mass blackout or a week-long shutdown of the financial system.

Hinton criticized Sam Altman, Mark Zuckerberg and Elon Musk, saying they understand the risks but would rather have the money. He called for FDA-style mandatory testing of frontier models before release, with independent evidence that they do more good than harm. He said he remains hopeful that catastrophe can be averted.

Why it matters: One of AI's founding figures is now arguing that the technology's risks may outweigh its benefits, a message that lands as labs race to deploy agents.

• Source: AI 'godfather' Geoffrey Hinton: it might turn out we were better off without AI — The Guardian
• Source: Geoffrey Hinton warns humanity could be better off without AI and predicts possible large-scale incident — AI Understanding

AI Security: 5 Stories to Know Today

11. WorkOS Says ARTEX AI Breached Internal Sales App

WorkOS said an attacker accessed an internal sales application on October 30, exposing business contact details and summaries of customer communications. The identity platform said no production services or customer application data were affected.

Founder Michael Grinich attributed the intrusion to ARTEX, an open-source AI penetration-testing framework, citing user agents "artex-enrich/1.0" and "norma/0.4." WorkOS disabled external access and hired an outside forensics firm. The company has more than 3,000 customers.

Why it matters: The breach shows an AI agent framework automating reconnaissance inside a company that sells identity infrastructure to other developers.

• Source: October 30th, 2026 Data Security Incident — WorkOS
• Source: WorkOS says attacker used ARTEX AI to breach internal sales app — BleepingComputer

12. Prompt Adware Found in OpenAI Codex Jailbreak Bundle

Check Point Research found a Codex jailbreak bundle called Haiou that rewrites the agent's global instructions so ordinary greetings trigger an advertisement for a third-party LLM gateway. The researchers coined the term "prompt adware" for the technique.

The bundle uses role-play and capture-the-flag framing to bypass guardrails, then injects persistent ads for the jailbreak author's LLM gateway. Check Point found no evidence the gateway is malicious or that data was stolen. The repository had 403 stars and 77 forks.

Why it matters: Agent configuration files are now an advertising and influence surface, and they can be changed without any executable payload for security tools to catch.

• Source: Prompt Adware: How Hidden AI Instructions Could Create a New Enterprise Security Risk — Check Point
• Source: Researchers Uncover Prompt Adware Hidden in OpenAI Codex Jailbreak Toolkit — The Hacker News

13. Anthropic's FlashProve Verifies Which Model Wrote a Response

Anthropic released FlashProve, an open-source prototype that uses zkSNARK proofs to verify that a specific model generated an output without revealing its weights. The system creates a fingerprint of the model and attaches a compact proof to each response.

It generates proofs at about 110 tokens per second on one H100 for Llama 4 Maverick 400B, with a 19MB proof that verifies in 17 seconds on a CPU. Overhead is 70x to 95x, and it only handles 1,024-token contexts and most-likely-token sampling.

Why it matters: Model identity is becoming a trust problem for routing services and regulators, though FlashProve's context and sampling limits keep it in the lab for now.

• Source: Anthropic is working on a system to prove AI models' identity — TechCrunch
• Source: Anthropic's new prototype proves which model wrote an AI response without exposing its weights — The Decoder

14. 3AM Ransomware Uses AI-Like Loop in UK School Attack

A 3AM ransomware affiliate encrypted about 37 of nearly 50 machines at a UK school estate, Bitdefender reported. The attacker reached the estate from an unmanaged host, built trusts between three Active Directory domains and deployed a Rust-based encryptor.

Bitdefender said the commands showed signs of an agentic loop: numbered script revisions, self-labeled output and retries spaced 40 to 100 seconds apart. It could not identify the model or confirm whether a human copied commands between the implant and an AI.

Why it matters: The case suggests attackers are testing a hybrid model where a person picks targets and a model handles repetitive troubleshooting, making some attacks noisier but faster.

• Source: UK school estate hit by 3AM ransomware in suspected AI-assisted attack — BleepingComputer
• Source: 3AM Affiliate Embraces AI-Like Loops in European School Attack — Dark Reading

15. CrowdStrike Adds an AI Kill Switch for Rogue Agents

CrowdStrike launched an AI Kill Switch in Falcon Guardian that lets an administrator terminate a misbehaving agent, block it from relaunching and cut its enterprise network access. The feature is available now.

The kill switch acts first at the AI gateway and will extend to endpoints later. CrowdStrike also made its AI Gateway generally available and added agent-level hooks for Claude Code, OpenAI Codex, GitHub Copilot and Cursor.

Why it matters: Enterprises want a human override before they give agents system-level access, and gateways are becoming the choke point for that control.

• Source: CrowdStrike Introduces the AI Kill Switch for AI Safety — CrowdStrike
• Source: CrowdStrike launches AI kill switch that it says can stop internal agents that go rogue — CNBC

More AI Stories Worth Reading

• Data center builders are turning to retired jet engines as grid connections and new turbines take years to arrive — TechCrunch
• Google's Search revenue growth slowed to 12% in Q3 even as queries grew at the fastest rate in five years — The Verge
• Ukrainian drone strikes are pushing Russia to plan underground data centers for its exposed digital infrastructure — Ars Technica
• Compute consumed 58% of Anthropic's revenue in Q3, up from 48%, as operating losses grew ahead of its IPO — TechCrunch

Related Reading on Kill The AI

• Google tied school Search to AI; OpenAI bought Pierre; data centers lost the midterms.
• NYT sued Brave over AI scraping; Anthropic's S-1 showed a $23.29B loss; SpaceX cut Grok limits.
• OpenAI capped paid accounts and merged Chat with Work; banks blocked AI shopping agents.
• Our data review of the AI boom's first eight weeks, from capex and chips to layoffs.

Methodology & Sources

Compiled November 5, 2026, using multi-source research. Outlets: CNBC, TechCrunch, The Verge, Wired, OpenAI, AMD, The Guardian, Business Insider, Businesskorea and Yahoo Finance. Also freegardner, Nvidia, The Decoder, SETI Institute, Tech Insider, AI Understanding, WorkOS, BleepingComputer, Check Point, The Hacker News, Dark Reading, CrowdStrike and Ars Technica. All sources are free to read; no paywalled links. Details are as reported at compilation time.

Frequently asked questions