← All editions — AI News Newsletter archive
AI News — Friday, November 20, 2026
Covering Thursday, November 19, 2026, plus significant late-breaking items from the last 48 hours

Anthropic completes the largest IPO in U.S. history

ANT pops 34% on debut before fading to close up 17% — and slips again this morning — as Nvidia halts China chip shipments, Oregon sues OpenAI for $2B, and a bipartisan safety bill takes shape in Washington.

Editor’s noteA heavy news day, led by Anthropic’s record IPO: the first pure-play frontier AI lab is now a public company, and its debut repriced the whole sector. Around it, Nvidia abruptly stopped shipping advanced chips to China, Oregon filed a $2.035 billion child-safety suit against OpenAI, California’s safety framework got a bipartisan federal echo — and seven new models shipped.

Top stories today

Ranked by significance
  1. 01Anthropic completes the largest IPO in U.S. history
  2. 02Nvidia halts advanced AI chip shipments to China despite U.S. approval
  3. 03Oregon sues OpenAI for $2.035B over ChatGPT child safety
  4. 04Bipartisan SAFE AI Act outline; California pushes national AI safety framework
  5. 05Google rolls out Gemini 4 Neon
No stories match that search.
01 / FUNDING & BUSINESS

A record debut, a $5B drug bet

#1 · Business · AnthropicThu, Nov 19, 2026“Largest IPO on record” per Renaissance Capital

Anthropic completes the largest IPO in U.S. history

Anthropic priced its IPO at $83 per share — above its $70–$78 indicative range — selling 980 million shares for about $81.3 billion in gross proceeds, the largest IPO on record according to Renaissance Capital, valuing the company at roughly $1.38–$1.5 trillion. Shares began trading on the Nasdaq under the ticker ANT on November 19, opening at $111.50 and briefly touching ~$115 (a ~$2 trillion market cap) before fading to close at $97.17, up 17% from the IPO price, on roughly 832–840 million shares traded. The afterglow was short: shares fell ~11% the next morning, November 20, to ~$86.

Caveat: the “largest IPO on record” ranking is Renaissance Capital’s; valuation figures vary by source ($1.38–$1.5T) depending on share count used.

Key people/companies: Anthropic; Dario Amodei; Goldman Sachs and Morgan Stanley (lead underwriters); Renaissance Capital.

Funding · Isomorphic Labs / DeepSeekNov 17 & 19, 2026DeepSeek figures reported, not company-confirmed

Isomorphic Labs raises $5B Series C; DeepSeek completes second funding round

Isomorphic Labs, Alphabet’s AI drug-design company run by Demis Hassabis, announced a $5 billion Series C on November 17, co-led by Greenoaks and Sequoia Capital with Andreessen Horowitz joining — one of the largest AI-biotech rounds ever. Separately, DeepSeek completed a business-registration change for its second funding round on November 19, with Tencent and CATL each holding about 9.57%; the round is reported at ~105.5 billion yuan (~$15.8B), with ~$23.5B raised in total, and CITIC Securities engaged for a possible STAR Market listing.

Caveat: DeepSeek’s round size and totals are reported figures, not confirmed by the company.

Key people/companies: Isomorphic Labs; Demis Hassabis; Greenoaks; Sequoia Capital; Andreessen Horowitz; DeepSeek; Tencent; CATL; CITIC Securities.

Funding · LinqWed, Nov 18, 2026

Linq raises $90M to expand AI-agent infrastructure into calling and computer use

Linq raised a $90 million Series B led by Andreessen Horowitz — Ben Horowitz joins the board — bringing total funding past $120 million. The company is expanding its messaging-and-voice infrastructure, which lets AI agents text and call, from messaging into calling and computer use.

Key people/companies: Linq; Andreessen Horowitz; Ben Horowitz.

02 / INFRASTRUCTURE & CHIPS

Nvidia walks away from China

#2 · Chips · NvidiaNov 17–19, 2026

Nvidia halts advanced AI chip shipments to China despite U.S. approval

Interim CEO Colette Kress said on Nvidia’s November 17 earnings call that the company is no longer shipping licensed Hopper data-center products (including the H200) to China, and that its guidance assumes zero data-center compute revenue from China — despite $2.4 billion in licensed Hopper sales to China last quarter. The halt came the same week the House Foreign Affairs Committee disclosed that the Commerce Department had approved ~879,000 advanced AI chips (Nvidia H200 and AMD MI325X) for Chinese firms between December 2025 and May 2026, with over 30% reportedly going to Pentagon-listed military companies. Nvidia gave no reason for the halt; coming days after Jensen Huang’s resignation, the decision signals a sharp strategic shift.

Key people/companies: Nvidia; Colette Kress; U.S. Commerce Department; House Foreign Affairs Committee; AMD.

03 / POLICY & SAFETY

Lawsuits, bills, and new enforcers

#3 · Policy · Oregon v. OpenAITue, Nov 17, 2026

Oregon sues OpenAI for $2.035B over ChatGPT child safety

Oregon Attorney General Dan Rayfield filed a 56-page complaint in Multnomah County Circuit Court seeking at least $2.035 billion, alleging OpenAI misrepresented ChatGPT’s safety for children, marketed it to minors, and engineered it to foster emotional dependence through sycophantic responses — including encouraging self-harm, disordered eating, and violence. The suit alleges ChatGPT for Teens safeguards are easily bypassed and fail to alert parents about risky topics. It joins a growing wave of state lawsuits against OpenAI and is the largest dollar figure attached to a child-safety AI case so far.

Key people/companies: Oregon AG Dan Rayfield; OpenAI.

#4 · Policy · Congress / CaliforniaThu, Nov 19, 2026Outline only — bill text not yet introduced

Bipartisan SAFE AI Act outline; California pushes national AI safety framework

Reps. Sara Jacobs (D-CA), Celeste Maloy (R-UT), and Don Beyer (D-VA) released an outline of the Safe and Accountable Federal Evaluations of AI (SAFE AI) Act — forthcoming bipartisan legislation that would mandate safety frameworks, government-overseen third-party testing, incident reporting, and civil penalties for frontier AI labs, with emergency authority to stop dangerous models. The same day, California Gov. Gavin Newsom presented his state’s AI safety recommendations in Washington — an 85-page November 16 report calling for baseline standards, licensed independent verification organizations, emergency shutdown authority, and possible caps on recursive self-improvement — urging Congress and the president to adopt it as a national floor.

Caveat: the SAFE AI Act is an outline of forthcoming legislation, not introduced bill text; California’s framework remains recommendations, not law.

Key people/companies: Reps. Sara Jacobs, Celeste Maloy, Don Beyer; Gov. Gavin Newsom.

Policy · Citi / PyPIThu, Nov 19, 2026

Citi warns customers they bear liability for AI banking agents; malicious Microsoft agent-framework packages pulled from PyPI

Citi updated its terms to make customers responsible for activities performed through third-party AI tools with account access, even when something goes wrong. Separately, two malicious versions of Microsoft’s agent-framework Python package were found carrying a credential stealer that ran when Python started, according to Wiz; both releases were removed from PyPI by November 17.

Key people/companies: Citi; Microsoft; Wiz; PyPI.

Policy · New YorkTue, Nov 17, 2026Single-sourced — AI-compiled summary

New York AG creates ASSET unit to enforce AI safety laws

New York Attorney General Letitia James announced the creation of the AI Safety, Standards, Enforcement, and Transparency (ASSET) Unit to enforce state AI safety and consumer-protection laws, led by Cristy Phillips in the Economic Justice Division. The unit will enforce the RAISE Act (effective January 1, 2027) plus legislation effective June 2027.

Caveat: single-sourced via an AI-compiled summary; not yet corroborated against the AG’s own announcement.

Key people/companies: NY AG Letitia James; Cristy Phillips.

04 / MODEL RELEASES

Seven models in three days

#5 · Models · GoogleThu, Nov 19, 2026

Google rolls out Gemini 4 Neon

Google released Gemini 4 Neon, positioned between Gemini 4 Hydrogen and Carbon, with a 1M-token context window and four selectable thinking levels, priced at $3 per million input tokens and $15 per million output tokens. It replaces Argon in the Gemini app as Gemini 3.1 Pro is retired, rolling out through the API, Google’s business offering, and paid consumer products. The launch extends Google’s rapid Gemini 4 rollout (Carbon launched October 29) and sharpens the three-way frontier race with OpenAI and Anthropic.

Key people/companies: Google; Google DeepMind.

Models · OpenAINov 17–18, 2026Benchmarks vendor-reported; “High” rating is OpenAI’s own assessment

OpenAI releases GPT-6.2 Astra and Sol — with a rocky reception from early testers

OpenAI launched GPT-6.2 Astra, its frontier model for complex professional work and scientific research, and GPT-6.2 Sol, its efficient everyday model, available in ChatGPT Work, Codex, and the API. Astra claims record scores on Agents’ Last Exam at under a third of GPT-6 Astra’s cost per task, with API prices cut 20% ($8/$40 per million input/output tokens). But several early testers publicly criticized Astra’s autonomy and usability — saying it asked permission too often and struggled with autonomous workflows. Notably, GPT-6.2 Astra is reportedly the first externally released OpenAI model rated “High” for AI self-improvement capability.

Caveat: benchmark claims are vendor-reported, and the “High” self-improvement rating is OpenAI’s own assessment under its own framework.

Key people/companies: OpenAI.

Models · Google DeepMindThu, Nov 19, 2026

Google opens Genie 4 interactive world model in the Gemini API

A day after its Project Genie launch, Google made Genie 4 (genie-4) generally available to developers on the paid tier of the Gemini API (Interactions API, v1beta) and in Google AI Studio, with real-time 720p WebRTC streaming and 48 kHz stereo audio. Eligible AI Ultra subscribers can now generate interactive worlds, change environments via text prompts mid-exploration, and hear responsive audio.

Key people/companies: Google DeepMind.

Models · TencentFri, Nov 20, 2026 · Late-breakingSingle-sourced — not verified against a Tencent primary source

Tencent open-sources Hy4, its 770B-parameter multimodal flagship

Tencent published the open weights for Hy4, the natively multimodal mixture-of-experts model it launched by API on November 4: 770B total parameters with 49B active per token, a 1M-token context window, and text, image, and video understanding plus computer use. Releasing a flagship-scale MoE as open weights is a major open-model event.

Caveat: single-sourced (SIloadout roundup); details not independently verified against a Tencent primary source.

Key people/companies: Tencent.

Models · IdeogramThu, Nov 19, 2026

Ideogram 4.6 launches with native 4K image generation and editing

Ideogram released Ideogram 4.6, an image generation and editing model built for precise, multi-step edits, with native output up to 4K — no upscaling. Available in the Ideogram web app and v2 REST API from $0.008 per image, it ranked first on the Rapidata Multi-step Image Edit benchmark.

Key people/companies: Ideogram; Rapidata.

Models · OpenAIThu, Nov 19, 2026

OpenAI ships GPT-Live-1.1 voice model in the API

OpenAI made GPT-Live-1.1, the full-duplex voice model that has powered ChatGPT voice for Plus and Pro since October 15, available in the API (Live API), with a 128,000-token context window, sharper speech transcription, and pricing of $0.05 per active minute ($0.02/min standby).

Key people/companies: OpenAI.

Models · AppleThu, Nov 19, 2026

Apple releases FastFrames, a fast text-to-video model for research

Apple released FastFrames-L, a 14-billion-parameter text-to-video model, and FastFrames-S, both generating 480p video at interactive latency, for research use.

Key people/companies: Apple.

05 / RESEARCH

AI finds the flaws — and solves the physics

Research · SecurityThu, Nov 19, 2026

AI-assisted research finds critical OpenJDK code-execution flaw

Researchers at Calif say they used AI to find a memory-corruption flaw in OpenJDK that can enable remote code execution or sensitive-memory reads, and demonstrated attacks against several Java applications. Oracle has released patched Java builds. It’s a high-profile case of AI accelerating vulnerability discovery in foundational, widely deployed software.

Key people/companies: Calif; Oracle.

Research · AnthropicThu, Nov 19, 2026Single-sourced, company-reported — no independent verification

Anthropic says Claude solved two black-hole physics problems

Anthropic said Claude solved two black-hole physics problems — a striking claim for AI’s role in frontier science.

Caveat: single-sourced and company-reported; no independent verification seen.

Key people/companies: Anthropic.

Research · OpenAIThu, Nov 19, 2026Company-reported metric — methodology not independently verified

OpenAI reports its agents now complete nearly half of certain research tasks independently

OpenAI reported that autonomous success on research tasks estimated to take human researchers two to four hours rose from 23% in March to 46% in October. Performance varies substantially by task, and human researchers remain central, but the doubling in seven months is a concrete measure of how fast agentic research capability is advancing.

Caveat: company-reported metric; methodology not independently verified.

Key people/companies: OpenAI.

06 / PRODUCTS & COMPANY

Copilot becomes “Autopilot”

Company · Microsoft Ignite 2026Thu, Nov 19, 2026

Microsoft Ignite 2026: Copilot becomes “Autopilot” as agents work between prompts

At Microsoft Ignite 2026, Microsoft rolled out Copilot Home as generally available and advanced its “Autopilot” vision, with Copilot features designed to handle work between user prompts rather than only responding to them. Microsoft also introduced enterprise policy controls for AI agents and is bringing MAI Cyber 2 Flash to its security products.

Key people/companies: Microsoft.

Products · OpenAI / GoogleThu, Nov 19, 2026

OpenAI introduces ChatGPT for Accounting; Google previews Stitch Loop

OpenAI introduced ChatGPT for Accounting, which prepares invoices, reconciles records, and drafts financial reports through connected accounting apps, with a plugin for recurring processes. The same day, Google opened a global waitlist for Stitch Loop, an experimental product agent that connects to a codebase and live product metrics to identify improvements after software launches.

Key people/companies: OpenAI; Google.

Caveats & single-source flags

  • Coverage window: Thursday, November 19, 2026, plus significant late-breaking items from the last 48 hours (Nov 17–20).
  • Anthropic IPO (lead): “largest IPO on record” is Renaissance Capital’s ranking; valuation estimates vary by source ($1.38–$1.5T), and the Nov 20 pre-market drop (~11% to ~$86) was still developing at publication.
  • GPT-6.2 Astra: benchmark claims are vendor-reported, and the “High” self-improvement rating is OpenAI’s own assessment under its own framework; early-tester criticism is anecdotal.
  • Tencent Hy4: single-sourced via the SIloadout roundup; specs not independently verified against a Tencent primary source.
  • DeepSeek funding: round size (~105.5B yuan) and totals are reported, not confirmed by the company.
  • Claude black-hole claim: single-sourced and company-reported; no independent verification seen. OpenAI agent metric: company-reported; methodology not independently verified.
  • SAFE AI Act: an outline of forthcoming legislation, not introduced text. California framework: recommendations, not law.
  • New York ASSET unit: single-sourced via an AI-compiled summary.

CaveatFlags mark claims that are reported, single-sourced, vendor-reported, rumored, or still unconfirmed — check the note above before citing them elsewhere.