AI News Today (Nov 14, 2026): Nvidia Fallout, ChatGPT Wallet
Last updated: Nov 14, 2026 — next refresh daily.
AI news today, November 14, 2026: Nvidia's Jensen Huang will testify before Congress as Commerce Secretary Howard Lutnick faces pressure. OpenAI launched ChatGPT Wallet, and Anthropic's IPO is six times oversubscribed. Here are the 10 top stories and 5 security stories, each with free, non-paywalled sources.
Key Takeaways
- Huang agreed to testify before Congress; Lutnick will face Senate questions on chip exports.
- OpenAI launched ChatGPT Wallet, letting dots agents buy items with user approval.
- Anthropic's IPO is six times oversubscribed, with 825 million shares at $70–$78.
- New York City banned surveillance pricing, with fines up to $1,000 per violation.
- The FBI says North Korean deepfake calls stole over $400 million in crypto.
1. Update: Nvidia Fallout Hits Lutnick as Huang Agrees to Testify
Jensen Huang agreed to testify before Congress by Thanksgiving after lawmakers threatened a subpoena. Commerce Secretary Howard Lutnick will testify before the Senate Banking Committee next month. Treasury Secretary Scott Bessent said several agencies had favored adding Megaspeed to the entity list for months.
Nvidia shares are down 11% this week and headed for their sixth straight weekly decline, the longest losing streak since 2012. The chipmaker has lost roughly $1.6 trillion in market value since early October. We first covered Huang's resignation in our November 13 briefing.
Why it matters: The scandal has moved from a personnel crisis to a referendum on US chip export controls, with Nvidia facing Congress and the market at once.
- Source: Jensen Huang agrees to testify before Congress — TechCrunch
- Source: Nvidia on track for longest weekly losing streak since 2012 as market value sinks by $1.6 trillion — CNBC
- Source: Pressure grows on Trump's Commerce Secretary Lutnick in fallout from Nvidia scandal — BBC
2. OpenAI Launches ChatGPT Wallet for Agent Purchases
OpenAI launched ChatGPT Wallet, a payment feature that lets its dots agents buy things for users. It is available first to US ChatGPT Pro subscribers over 18, and works with American Express, Mastercard and Visa cards.
Users add a card once and approve each purchase with Face ID or a fingerprint. The payment network issues a token, so the agent and merchant never see the actual card details. Rewards and dispute protections stay with the card. OpenAI says merchants receive shipping and contact details automatically.
Why it matters: Agentic commerce now has a checkout rail from the company that already helps 1.2 billion people a week research what to buy.
- Source: Introducing ChatGPT Wallet — OpenAI
- Source: OpenAI's ChatGPT Wallet allows 'dots' AI agents to make purchases — CNBC
- Source: ChatGPT's new Wallet lets AI agents shop and check out for you — TechCrunch
3. Update: Anthropic IPO Is Six Times Oversubscribed
Anthropic's IPO is about six times oversubscribed, with orders exceeding available shares, according to Bloomberg. The company plans to sell 825 million shares at $70 to $78 each, seeking up to $90.4 billion including a $26 billion private placement.
At the top of the range, Anthropic's fully diluted valuation would be about $1.4 trillion, below the $2 trillion some investors once expected. Existing backers get priority in allocations. We first covered the IPO terms in our November 11 briefing.
Why it matters: Demand that deep suggests investors are still paying up for frontier AI, even as the tech trade wobbles and the valuation falls short of earlier hopes.
- Source: Anthropic's IPO is six times oversubscribed and on track to become the largest public offering in history — The Decoder
- Source: Anthropic is Seeking to Raise Over $90.4 Billion. The IPO is Reportedly Oversubscribed Multiple Times. — Yahoo Finance
4. Meta Muse Starts Calling Businesses for Users
Meta's Muse agent can now place calls to US businesses for paying subscribers, handling tasks like cable companies, the DMV and restaurant reservations. Meta AI chief Alexandr Wang says Muse has more than 10 million users.
The feature was tested in September and is available to all paying Muse subscribers in the US. Many businesses hang up when they detect an AI voice, which forced Meta to use human operators in some tests.
Why it matters: Voice calls are one of the last interfaces AI agents have not automated, and Meta is testing whether businesses will treat an AI caller like a customer.
- Source: Meta's Muse now calls businesses for you, but many companies might just hang up on the AI — The Decoder
- Source: Muse can now place calls to US businesses — if you pay — TechCrunch
5. New York City Bans Surveillance Pricing
New York City's council passed a bill banning surveillance pricing, which uses personal data and algorithms to charge different customers different prices for the same goods or services. The ban covers online and in-store sales and allows bona fide discounts. The bill applies to retailers and other businesses.
The bill passed 44-5 and has Mayor Zohran Mamdani's support. The Department of Consumer and Worker Protection can fine companies up to $1,000 per violation. New York is the second US city to pass such a ban, after Seattle.
Why it matters: The vote adds a local enforcement layer to a pricing practice that is largely invisible to shoppers, and it gives other cities a template to copy.
- Source: NYC to bar surveillance pricing as momentum for bans grows nationwide — The Record
- Source: New York City Council Approves Landmark Legislation Banning Surveillance Pricing — New York City Council
6. FAA Expands AI Traffic Tool to All 22 Centers
The FAA expanded its SMART AI-assisted traffic planning tool to all 22 US en route centers after a pilot in the Washington area. The system combines about 200 data streams, including weather, flight paths and controller staffing, to recommend more efficient routes.
The FAA says SMART helped reroute 180 flights, saving an average of 10 minutes and nearly 80 gallons of fuel per flight. FAA specialists review the recommendations before passing them to controllers, who decide whether to accept them. The controllers' union says it was not involved in designing or testing the system.
Why it matters: The expansion puts AI advice into every corner of US airspace, testing whether machine recommendations can reduce delays without eroding controller trust.
- Source: FAA expands SMART air traffic tool to all 22 US en route centers — AeroTime
- Source: FAA expands AI advice to manage air traffic congestion across entire US airspace — Ars Technica
7. Anthropic Launches Claude for Manufacturing
Anthropic launched Claude for Manufacturing, a plugin with 17 skills for quoting, quality investigations, maintenance and purchasing. It connects to Autodesk Fusion, Infor, Microsoft 365 and Oracle NetSuite, and drafts quotes and reports from a manufacturer's records.
The workflows keep people in the loop: staff review and approve quotes, corrective actions and supplier orders before anything is sent. Infor also released its own plugin that lets authorized users make changes in its ERP system, subject to their existing permissions. The plugin is available to Team and Enterprise customers.
Why it matters: Manufacturing is full of structured records and repetitive paperwork, making it a strong test of whether enterprise agents can save time without controlling the factory floor.
- Source: Anthropic's Claude for Manufacturing plugs AI directly into factory software to automate industrial workflows — The Decoder
- Source: Infor and Anthropic Announce Partnership That Turns Insights into Actions for Manufacturing Operations — Infor
8. LangChain Raises $300M at $3B Valuation
LangChain raised $300 million at a $3 billion valuation in a round led by Alphabet's CapitalG. The company makes open-source tools and the LangSmith platform for building, monitoring and deploying AI agents.
The valuation is more than double the $1.25 billion LangChain set in October 2025. IVP, Sapphire Ventures, GIC and DFJ Growth also participated. LangChain says 7,000 customers use LangSmith, including Nvidia, Toyota and Salesforce. The company plans to reach about 850 employees by the end of 2027. It has not disclosed revenue.
Why it matters: Agent infrastructure is becoming its own software market, and LangChain's raise shows investors expect enterprises to need plumbing between models and business systems.
- Source: AI agent startup LangChain hits $3B valuation — TechCrunch
- Source: LangChain raises $300 million at a $3 billion valuation in a round led by Alphabet's CapitalG — MLQ.ai
9. AI Helps Unlock 663 Theorems in Gödel's Notebooks
A Helsinki logician used Anthropic's Claude to analyze 1,009 pages of Kurt Gödel's notebooks, written in an obsolete German shorthand. Claude produced a digest in days and identified 663 theorems with a proof, proof sketch or proof idea.
A proof assistant called Coq then checked the results: about 400 were confirmed as stated, about 200 needed clarification or extra assumptions, and about 50 turned out to be false. The work was reported by The Guardian.
Why it matters: The project shows AI can help historians of mathematics surface forgotten results, while also showing why machine-generated proofs still need human and formal verification.
- Source: 'Ahead of his time': AI helps unlock discoveries hidden in notebooks of man who transformed maths — The Guardian
- Source: AI helped uncover hundreds of unknown theorems in Gödel's notes — UA.NEWS
10. OpenAI Board Member Warns of AI Takeover Risk
Paul Christiano, a board member at the OpenAI Foundation and its safety committee, warned that AI could seize control within 6 to 12 months with meaningful probability. He said that if labs race ahead without checks, the three-year risk looks like a coin flip.
Christiano said no quantitative model can put a number on that risk, but he argued that current policy responses are not on track to work. He estimated a 4% one-year risk and 15% over three years in September. He suggested the industry may need to slow down or stop to build stronger protections.
Why it matters: A safety board member publicly describing takeover risk in months, not decades, raises the pressure on labs to prove their controls before the next model launch.
- Source: OpenAI board member worries an AI 'takeover' of society could be months away — Ars Technica
- Source: OpenAI's Paul Christiano warned AI takeover could come within months — Quartz
AI Security: 5 Stories to Know Today
11. FBI Says Deepfake Calls Stole $400 Million in Crypto
The FBI said North Korean hackers stole more than $400 million in cryptocurrency over 18 months using a scheme called FaceStealer. Attackers hijacked social media accounts, invited victims to fake Zoom, Teams or Google Meet calls, and used AI-generated deepfakes to impersonate contacts.
The hackers secretly recorded victims and reused the footage to target their contacts. The FBI has seized more than 900 victim webcam videos and says the group targets crypto and Web3 workers, especially open-source developers.
Why it matters: Deepfakes make the oldest social-engineering trick, a call from someone you trust, scale across thousands of targets with minimal manual work.
- Source: FBI Says North Korean Hackers Stole Over $400 Million in Crypto Using Deepfake Video Calls — The Hacker News
- Source: North Korean hackers used fake video calls in $400m crypto hack, says FBI — BBC
12. Mior and Skyfall Break AMD Confidential VMs
Researchers at ETH Zürich disclosed Mior and Skyfall, two attacks that let a malicious cloud host read or modify data inside AMD SEV-SNP confidential virtual machines. SEV-SNP is designed to keep a VM's memory encrypted even from the hypervisor.
AMD assigned Mior CVE-2026-0462 and Skyfall's VM leak CVE-2026-40671, and released firmware and microcode fixes in August. Platform firmware for Skyfall is due in November and for Mior in December. Neither advisory says the flaws have been exploited.
Why it matters: Confidential computing is a selling point for AI clouds, and a hypervisor-level flaw undermines the promise that sensitive workloads stay private from the host.
- Source: New Mior and Skyfall Attacks Let Rogue Cloud Hosts Steal Secrets From AMD SEV-SNP VMs — The Hacker News
- Source: Mior: Breaking AMD SEV-SNP with Device-as-Memory — XCA
13. Trezor Finds Implants in Three Hardware Wallets
Trezor confirmed that three Safe 5 hardware wallets sold by Southeast Asian reseller CryptoBilis contained malicious implants. The extra circuit board reads the recovery phrase from the wallet's screen and sends it to attackers over a cellular network.
Trezor released firmware version 2.13.1, which stops the implants found so far from capturing anything on wallets that run it. The company says the update cannot protect a backup that was captured before it, and has no reports of lost funds.
Why it matters: A hardware wallet is only as trustworthy as its supply chain, and a compromised reseller can defeat the offline storage model before the device is ever used.
- Source: Trezor confirms crypto-stealing implants in three wallets, releases fix — BleepingComputer
- Source: Trezor Confirms Hidden Implants in CryptoBilis-Sold Wallets That Can Steal Recovery Phrases — The Hacker News
14. Check Point Gateways Wiped via PHP2Shell
Attackers exploited PHP2Shell flaws in unpatched Check Point Security Gateways to deploy a backdoor called TSGA and a wiper that destroyed VMware ESXi files. The campaign also used a previously unreported local privilege escalation flaw in Gaia OS.
Check Point released emergency hotfixes for a zero-day that let attackers gain root and persist since late October. The PHP2Shell bugs were disclosed earlier this year and have been used in attacks on other internet-facing systems.
Why it matters: Edge appliances remain a favorite entry point, and a wiper that targets ESXi can turn a gateway compromise into a full virtualization outage.
- Source: Check Point gateways hacked via PHP2Shell to deploy VMware ESXi wiper — BleepingComputer
- Source: Hackers Exploit PHP2Shell Flaws in Check Point Gateways to Deploy TSGA Backdoor and Wipe ESXi Files — The Hacker News
15. OpenAI Launches Daybreak Red Team and Defense Factory
OpenAI launched Daybreak Red Team and Daybreak Defense Factory, two products that use AI agents to automate security testing and patching. Red Team probes web applications, validates scanner and bug-bounty findings, and tests fixes.
Defense Factory scans connected GitHub repositories, investigates findings and prepares draft pull requests with patches and tests. Engineers review every change, and patches are not deployed automatically. OpenAI has not published pricing.
Why it matters: AI is now automating both sides of the cyber arms race, and the guardrails around patch generation will determine whether defenders gain more than attackers.
- Source: OpenAI enters the cybersecurity market with two Daybreak products — The Decoder
- Source: OpenAI launches Daybreak Red Team and Defense Factory, AI agents that pen-test web apps and draft security patches — MLQ.ai
More AI Stories Worth Reading
- Microsoft's global outage knocked out Outlook, Teams and Copilot in 74 countries for more than 12 hours. — Al Jazeera
- ElevenLabs launched ElevenAI, a preview app that combines meeting notes, dictation and agent follow-ups for teams. — Digital Trends
- Oura will acquire AI-native primary care startup Counsel Health for $625 million in cash. — Counsel Health
Related Reading on Kill The AI
- Nvidia's CEO resignation, Gemini 4 Hydrogen, and Nebius's $40 billion plan.
- OpenAI's $10 billion quarter, TSMC's AI bet, and AlphaGenome 2.
- Anthropic's IPO terms, Nvidia's Wachtell probe, and Gemini 4 Carbon.
- Our data review of the AI boom's first eight weeks, from capex to layoffs.
Methodology & Sources
Compiled November 14, 2026, using multi-source research. Outlets: TechCrunch, CNBC, BBC, OpenAI, The Decoder, Yahoo Finance, The Record, New York City Council, AeroTime, Ars Technica, Infor, MLQ.ai and The Guardian. Also UA.NEWS, Quartz, The Hacker News, XCA, BleepingComputer, Al Jazeera, Digital Trends and Counsel Health. All sources are free to read; no paywalled links. Details are as reported at compilation time.
Frequently asked questions
Nvidia's Jensen Huang agreed to testify before Congress as Commerce Secretary Howard Lutnick faces pressure over chip exports. OpenAI launched ChatGPT Wallet, Anthropic's IPO is six times oversubscribed, and New York City banned surveillance pricing. The FBI also said North Korean deepfake calls stole more than $400 million in crypto.
ChatGPT Wallet is OpenAI's payment feature for its dots agents. It is available first to US ChatGPT Pro subscribers over 18 with American Express, Mastercard or Visa cards. Users approve each purchase with Face ID or a fingerprint, and the payment network issues a token so the agent and merchant never see the card details.
Commerce Secretary Howard Lutnick faces questions over why Megaspeed, the firm linked to the woman at the center of the Nvidia scandal, was not placed on the US entity list. Treasury Secretary Scott Bessent said several agencies had favored the move for months. Lutnick will testify before the Senate Banking Committee next month.
Anthropic's IPO is about six times oversubscribed, according to Bloomberg. The company plans to sell 825 million shares at $70 to $78 each, seeking up to $90.4 billion including a $26 billion private placement. At the top of the range, its fully diluted valuation would be about $1.4 trillion.
The FBI said North Korean hackers stole more than $400 million in cryptocurrency over 18 months using fake video calls and AI-generated deepfakes. The attackers hijacked social media accounts, impersonated contacts on fake Zoom, Teams or Google Meet calls, and secretly recorded victims. The FBI seized more than 900 victim webcam videos.
Last updated: Nov 14, 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.