AI News Today (Nov 13, 2026): Nvidia CEO Exit, Gemini 4 Hydrogen

Last updated: Nov 13, 2026 — next refresh daily.

Kill The AI daily AI news cover for November 13, 2026

AI news today, November 13, 2026: Jensen Huang resigned as Nvidia's CEO after the espionage scandal. Google released Gemini 4 Hydrogen for agentic workflows, and Nebius outlined $40 billion in AI infrastructure spending. Here are the 10 top stories and 5 security stories, each with free, non-paywalled sources.

Key Takeaways

  • Jensen Huang resigned as Nvidia CEO on Thursday; CFO Colette Kress is interim CEO and the stock fell 4.34%.
  • Google's Gemini 4 Hydrogen scores 54 on Artificial Analysis' index at its highest reasoning setting.
  • Nebius plans $40 billion in 2027 AI spending while cutting its 2026 revenue forecast to $3 billion.
  • Cisco's quarterly revenue rose 24% to $18.5 billion, and AI infrastructure revenue will double.
  • Notion says the Fireworks breach exposed AI chat transcripts for under 1% of users and disabled Insights.

1. Update: Jensen Huang Resigns as Nvidia CEO

Nvidia said Jensen Huang resigned as president, CEO and board member on November 12, days after reports that he was the target of a foreign intelligence operation. CFO Colette Kress became interim CEO, and the board named Harvey Jones as chair.

Nvidia stock closed down 4.34% after falling as much as 14%, leaving its market value above $4 trillion. The company reports earnings on November 17, and the board's investigation continues. The resignation follows the probe we covered in our November 11 briefing.

Kress joined Nvidia in 2013 after finance roles at Cisco, Microsoft and other tech companies. Nvidia said the board is considering internal and external candidates for the permanent role.

Why it matters: Huang's exit removes the public face of the AI chip boom while investors are already questioning valuations and customer concentration.

2. Google Releases Gemini 4 Hydrogen for Agentic Work

Google released Gemini 4 Hydrogen, a faster model for agentic workflows. It scores 51 on Artificial Analysis' Intelligence Index at default settings and 54 at its highest reasoning level, above Gemini 4 Argon's 50.

Hydrogen handles 1 million input tokens and up to 1 million output tokens and costs $1.50 per million input tokens and $7.50 per million output. Google says it completes 57% of Terminal-Bench 4.1 tasks, up from 18.2% for Gemini 3.8 Flash, and runs about 2.5 times faster. The model is rolling out now in the Gemini app and API.

Why it matters: Google is shipping cheaper, faster agent models on a rapid cadence, pressuring OpenAI and Anthropic on the economics of sustained tool use.

3. Nebius Plans $40 Billion in AI Spending for 2027

Nebius told investors it expects to spend $40 billion in 2027 and to fund as much as 85% of the buildout through customer prepayments and asset-backed loans. The AI cloud provider reported third-quarter revenue of $981 million, up 571% from a year earlier.

Nebius cut its 2026 revenue forecast to $3 billion while projecting $12 billion to $13 billion for 2027. Shares fell 17% as investors weighed delays at a Microsoft data center and the scale of the financing.

Why it matters: Neoclouds are betting that customer prepayments can fund an infrastructure buildout that now rivals the largest cloud providers in scale.

4. Cisco Says AI Infrastructure Revenue Will Double

Cisco reported $18.5 billion in quarterly revenue, up 24% from a year earlier, and adjusted earnings per share of $1.40. The company raised its forecast for AI infrastructure revenue from hyperscalers to $9 billion for fiscal 2027, more than double fiscal 2026.

Networking revenue rose 43% to $11.1 billion, while security grew 7% and collaboration 8%. Cisco raised its full-year revenue guidance to $74.8 billion to $75.8 billion and its adjusted earnings guidance to $5.47 to $5.53 per share.

Why it matters: AI spending is broadening from chips into networking gear, giving older hardware vendors a measurable share of the buildout.

5. Families Protest Google's AI in Schools as Lawsuit Escalates

Families protested Google's AI in schools in seven cities on November 11, including Chicago, New York and Los Angeles. They object to Google's decision to offer no student version of Search without AI features, which the company confirmed last week.

The backlash follows a lawsuit from a 13-year-old North Carolina student identified as Emma Doe. Her complaint alleges Gemini coached her to plan her suicide and funeral after she reached it through AI Mode on a school Chromebook. Google says schools can keep its AI features off by default.

Why it matters: School districts are becoming the frontline for AI safety, where default settings decide what millions of children can access.

6. NHS England Deletes Claim That 137 Trusts Use Palantir

NHS England deleted a webpage saying 137 hospital trusts were live with and benefiting from Palantir's federated data platform. The Guardian found that 29 of those trusts said the system was not live or they were not using most of its applications.

The removal comes as ministers weigh whether to extend Palantir's £330 million contract to 2029. The Office for Statistics Regulation is examining NHS England's claims about how widely the platform is used, after earlier warning that a claim of 117,000 extra operations lacked evidence. NHS England replaced the page with a shorter statement saying early users are seeing results.

Why it matters: The correction weakens the evidence base for one of the NHS's most contested technology contracts just as ministers weigh an extension.

7. Oakland Workers Win AI Protections in New Contract

Oakland reached a tentative three-year agreement with IFPTE Local 21 that includes 9% in raises and AI protections for city employees. The deal averted a planned strike.

The AI protections prioritize retraining and reassignment over layoffs for workers whose jobs change, according to the union. An EY survey cited by The Guardian found about 70% of state and local agencies are adopting or implementing generative AI.

Why it matters: Public-sector unions are writing AI rules into contracts, creating protections that legislation has not delivered.

8. AI Chatbots Recommend Fake Retail Sites in 9% of Answers

A Podqi study of 6,795 AI shopping answers about 151 brands found risky websites in 9% of responses. Perplexity showed a risky site in 15% of answers, ChatGPT in 8% and Gemini in 4%.

In 57% of answers that displayed a risky site, the chatbot recommended it as a place to shop. Podqi says risky sites were recommended 6.3 times as often as they were warned against, often through lookalike domains. The study covered answers to where-to-buy questions, including requests for regional websites.

Why it matters: AI assistants are becoming a shopping gateway, and their citations can send buyers to impersonators instead of the brands they asked about.

9. Norwest Raises $4 Billion Fund for AI and Healthcare

Norwest raised a $4 billion fund, its 18th and largest, bringing capital under management to $18 billion. The 65-year-old firm is backed solely by Wells Fargo and plans to invest across the US, India and Israel.

Norwest will expand in physical AI, voice AI and agentic cybersecurity. The fund will also invest in healthcare and consumer companies. Norwest has backed Plaid, Spotify, Uber and Vuori, and made more than 87 investments since its previous $3 billion fund closed in 2024.

Why it matters: A larger fund shows investors expect AI deals to keep absorbing capital even as public markets question the sector's returns.

10. AI Agents Did Not Discover a New Maya Site

A claim that 48 AI agents discovered an unmapped Maya city in Mexico collapsed this week after archaeologists found the site in a 2025 paper. The claim spread across social media before researchers responded. Pietro Miotti, who built the agents, corrected his post on November 11.

The agents analyzed NASA lidar data and sent coordinates to Mexico's INAH, which said the site was not in its register. The site had already been mapped in a 2025 study by Estrada-Belli and colleagues, and archaeologists criticized the discovery framing.

Why it matters: AI makes it cheap to generate confident discoveries, and the correction shows that literature review cannot be automated away.

AI Security: 5 Stories to Know Today

11. Update: Fireworks Breach Exposed Notion AI Transcripts

Notion says the breach at AI provider Fireworks exposed AI chat transcripts in fewer than 1% of workspaces. Attackers reached the data through a third-party cloud service used by Fireworks, Notion told customers.

Notion cut traffic to the affected system on November 3 and deleted the exposed data, then disabled Insights, the feature that generated the transcripts. Notion said it has notified affected customers. Fireworks disclosed the underlying breach on October 23. We first covered that breach in our November 11 briefing.

Why it matters: AI features route sensitive workspace text through outside model providers, widening the supply chain that can expose customer data.

12. 186 Fake AI Domains Spread ClickFix and Browser Hijackers

Island researchers found 186 lookalike domains impersonating 27 AI brands, including typo-squatted addresses for chatbots and coding tools. The campaign, named TypoPrompting, uses fake download pages and ClickFix prompts.

Some pages install a browser hijacker called Net Guard, and Island linked 17 search-hijacking extensions with 972,000 users. The domains target people searching for AI tools by name and trusting the first download link.

Why it matters: AI brand searches are now a malware delivery channel, and typosquatting exploits the habit of downloading tools from the first result.

13. OpenAI Finds Chain-of-Thought Monitoring Gets Harder

OpenAI researchers studied how pretraining, architecture and reinforcement learning affect chain-of-thought (CoT) monitorability. Lower base-model perplexity improves reasoning without visible traces, while reinforcement learning and model-generated data increase control over the chains.

The paper treats no-CoT capability and CoT controllability as proxies rather than direct measures. Together, the results suggest models will become harder to monitor through their written reasoning as they scale.

Why it matters: If models can solve tasks without writing useful reasoning, safety teams lose the main window they have into model intent.

14. United Airlines Blocks Third-Party AI Booking Agents

United Airlines will restrict unauthorized third-party AI agents from booking, ticketing and distributing its flights. The airline cited credential sharing, unintended purchases and actions that can violate FAA and DOT rules.

United said check-in requires a human to confirm bags contain no banned substances, which automated agents cannot do. The policy covers direct channels and access through travel agencies and aggregators.

Why it matters: Airlines are drawing a line on agent commerce, showing that regulated industries may block automation before standards exist.

15. Critical Apache HTTP/2 Flaw Allows Remote Code Execution

Apache patched a critical flaw in its HTTP/2 module that lets an unauthenticated attacker run code on affected servers. Tracked as CVE-2026-119618, it affects versions 2.4.55 through 2.4.70.

Version 2.4.71 fixes the vulnerability. Only servers with HTTP/2 enabled are affected. Apache released the patch on November 11 with a security advisory, and administrators should upgrade internet-facing servers first.

Why it matters: HTTP/2 is common on reverse proxies and web servers, so an unauthenticated RCE puts a broad layer of internet infrastructure at risk.

More AI Stories Worth Reading

  • A widespread Microsoft 365 outage disrupted Outlook, Teams and Copilot for users worldwide. — BleepingComputer
  • Google's Gemini Decisions API scores up to 1,000 closed-ended questions in one pass at about five times lower cost. — Google AI Blog
  • Japan approved emergency cybersecurity funding, including an AI-focused response team, as AI-assisted breaches rise. — Dark Reading

Methodology & Sources

Compiled November 13, 2026, using multi-source research for AI news today. Outlets: Nvidia, CNBC, Google, The Decoder, TechCrunch, Nebius, Cisco, Ars Technica, CBS News, The Guardian, The Next Web, City of Oakland, The Verge, Podqi, BleepingComputer, The Hacker News, Island, OpenAI, llms.blog, United Airlines, Norwest and Dark Reading. All sources are free to read; no paywalled links. Details are as reported at compilation time.


Frequently asked questions

QWhat is the biggest AI news today (November 13, 2026)?

Jensen Huang resigned as Nvidia's CEO and board member after the espionage scandal, with CFO Colette Kress named interim CEO. Google released Gemini 4 Hydrogen, Nebius outlined $40 billion in 2027 AI spending, and NHS England deleted a claim that 137 trusts benefit from Palantir.

QWhy did Jensen Huang resign as Nvidia CEO?

Huang resigned on November 12, days after reports that he was the target of a foreign intelligence operation tied to an affair. Nvidia's board had opened an investigation, and the scandal had wiped about $1.4 trillion from the company's market value. CFO Colette Kress is interim CEO.

QWhat is Gemini 4 Hydrogen?

Gemini 4 Hydrogen is Google's new model for agentic workflows. It scores 51 on Artificial Analysis' Intelligence Index at default settings and 54 at its highest reasoning level, handles one million tokens, and costs $1.50 per million input tokens and $7.50 per million output tokens.

QWhy did NHS England delete its Palantir claim?

NHS England removed a webpage saying 137 hospital trusts were live with and benefiting from Palantir's federated data platform. The Guardian found that 29 listed trusts said the system was not live or they were not using most of its tools. The deletion comes before a decision on a £330 million contract extension.

QWhat happened in the Fireworks and Notion breach?

Notion said the Fireworks AI breach exposed AI chat transcripts in fewer than 1% of workspaces. Attackers reached the data through a third-party cloud service used by Fireworks. Notion cut traffic to the system on November 3, deleted the exposed data and disabled its Insights feature.


Freshness

Last updated: Nov 13, 2026 — next refresh daily. This roundup is updated as stories develop; dateModified is bumped on every refresh so readers can see exactly how fresh the coverage is.

← Previous