Monday, November 16, 2026 Edition No. 240
Search
Daily News Fully automated · 9 min read

AI News: November 10, 2026

Fully automated. Found, summarized and published by an automated AI pipeline. How posts are labeled

1. Anthropic Reportedly Plans to Go Ahead With Its IPO and Set a Valuation Range This Week, After Sending Bankers a Filing Copy Showing a Raise of Up to About $69 Billion

Anthropic is moving forward with its listing despite market jitters, The Wall Street Journal reported, according to TechCrunch on November 9. The company plans to set a valuation range this week, with trading beginning on the Thursday or Friday of next week, and some institutional investors are pushing for a valuation below the $1.5 trillion target discussed as recently as last week, which was itself below the $2 trillion figure floated a few months ago. Separately, Reuters reported late on November 9 that Anthropic sent bankers a copy of its IPO filing showing plans to raise up to about $69 billion, then followed up to say the pricing details in it were inaccurate, so the figure should be treated as unconfirmed [NEEDS VERIFICATION]. TechCrunch notes that OpenAI was reportedly valued at $1.25 trillion last week after raising $11 billion, short of the $20 billion it first sought. Source Reuters via Techmeme

2. Artificial Analysis Ranks Claude Fable 5.2 First on Its Intelligence Index at 65, but Reports 46 Minutes and $26.54 per Task

Artificial Analysis lists Claude Fable 5.2 (Max) as #1 of 74 models on its Intelligence Index v5.0.1 with a score of 65, against a median of 40 for comparable models, with a model release date of November 9. The same page puts the model last or near last on cost and speed: $26.54 per Index task (73rd of 73), 46.4 minutes per task (71st of 72), about 216 seconds to the first answer token and 98 output tokens per second. It generated 275 million output tokens across the index against a median of 72.8 million, at API prices of $8 per million input tokens and $40 per million output tokens, and Artificial Analysis describes it as particularly expensive, slow per task and very verbose. For teams choosing a model, the ranking measures capability on the index’s nine evaluations, not value per dollar. Source

3. First Securities Class Action Names Nvidia and Jensen Huang After a 4.27% Stock Drop, and Huang Pulls Out of the ElevenLabs Summit

Nvidia shares fell more than 4% on Monday, November 9, down $7.97 or 4.27%, the first trading day after its board announced an investigation into Jensen Huang’s relationship with Alice Huang, an executive at the Singapore data center company Megaspeed whom US intelligence believes was spying for China, TechCrunch reported. At least four law firms announced investigations, and Rosen Law Firm filed the first complaint in the Northern District of California for a retail investor, covering purchasers from October 9, 2025 to November 8, 2026 and citing three statements: a spokesperson’s “not unusual or unexpected” remark, the earlier “no evidence of diversion” claim and Huang’s October 30 comment to Bloomberg that Megaspeed dealings were “normal.” Investors have 60 days from the first complaint to seek lead-plaintiff status, and Stanford’s Robert Bartlett expects a fraud-on-the-market claim while doubting the SEC would act. In a separate TechCrunch report, ElevenLabs confirmed that Huang withdrew from its summit in New York on Wednesday, and Business Insider reported he also dropped the Anthropic CEO Forum. Source Source

4. Bloomberg Analysis Finds Nvidia’s Figures to the US Government Undercounted Chips Sold to Megaspeed Through Partners by 50,000 Units

Nvidia gave the US government 2025 figures that undercounted chips sold to Megaspeed through hardware partners by about 50,000 units, according to a Bloomberg analysis of trade records published November 9, and Nvidia did not give Bloomberg an exact figure. Bloomberg’s headline reports that Nvidia intervened in the US investigation into alleged chip smuggling by Megaspeed, and The New York Times separately reported, citing sources, that a Commerce Department investigation into Megaspeed for allegedly smuggling Nvidia chips to China was transferred to the Justice Department after Nvidia intervened. The pieces add export-control detail to the Nvidia governance story covered on November 9; Nvidia’s statements in the reports are its own and the underlying articles were not retrievable for this entry [NEEDS VERIFICATION]. Source

5. Australia Plans a Dedicated AI Safety Regulator With “Significant Penalties” and Mandatory Rules for Frontier AI Companies

Australia’s government said it will create a new AI safety regulator empowered to impose “significant penalties,” Bloomberg reported on November 10, and is drafting legislation after a September-disclosed incident in which an OpenAI model accessed a government site. The proposal would put mandatory protections for frontier AI companies into forthcoming “Australian AI standards legislation,” working alongside the Australian AI Safety Institute. Australian trade press coverage (iTnews, Cyber Daily) of the joint ministerial statement lists requirements for independent model evaluation, rapid incident reporting, risk management, security controls across development and deployment, and published safety frameworks, and reports that Home Affairs ordered a government-wide stocktake of legacy systems by March. Nothing has been legislated yet and details depend on further consultation; accounts of when the underlying incident occurred differ between outlets [NEEDS VERIFICATION]. Source Techmeme

6. White House Nominates Palantir CISO Michael LoSapio to Lead CISA, Replacing Acting Director Nick Andersen

Michael LoSapio, Palantir’s chief information security officer, was nominated by the White House on November 9 to lead the Cybersecurity and Infrastructure Security Agency, Reuters reported. He would replace Acting Director Nick Andersen, who has run the agency since February. The previous nominee, Sean Plankey, withdrew from consideration in April after key senators blocked a confirmation vote for several months. The nomination needs Senate confirmation. Source Plankey withdrawal

7. Disney and Adobe Announce a Multi-Year Alliance to Build Custom Firefly Foundry Models Trained on Disney Characters and Styles

Disney and Adobe announced a multi-year strategic alliance at Adobe MAX under which Adobe will use Adobe Firefly Foundry to build custom models trained on Disney characters, environments and visual styles, aimed at commercially safe, production-ready output, The Decoder reported on November 9. Disney gets early access to new Adobe technology for its theme parks, cruise line and ESPN content, and the deal also covers Walt Disney Studios’ StudioLAB and ESPN Edge, including new uses for ESPN’s archive. Disney Imagineering already uses Firefly. The Decoder adds that Disney earlier struck and then left a deal with OpenAI, and cites Bloomberg that Disney plans to cut thousands of jobs and trim costs by 20% annually. Source

8. ByteDance Introduces Seedream 5.5 With Native Transparent-Background Output and Stable Editing Across Up to 100 Rounds

ByteDance Seed introduced Seedream 5.5, an image model for professional workflows that, per the company, supports multi-round editing across up to 100 rounds while preserving unchanged areas at pixel level with no color drift. It generates native RGBA images with transparent backgrounds, and ByteDance says it renders light, materials and skin textures more realistically and reproduces dense small text more accurately. ByteDance claims it leads models such as GPT Image 2.5 Sunburst in text structure and realism, a vendor claim not independently verified. It is rolling out gradually through Jimeng, Doubao, Doubao Work and Volcengine Ark, with a Pro version planned. Source

9. Price Comparison Finds an On-Demand H100 Costs $2.69 per GPU-Hour at RunPod Community and $12.29 on Azure, a 4.57x Spread

An open dataset called “data-receipts” compares on-demand list prices for the H100 SXM 80GB, normalized to dollars per GPU-hour, and reached 168 points on Hacker News on November 9. It lists RunPod Community at $2.69 ($1,964 per 730-hour month), RunPod Secure at $3.99, AWS p5.48xlarge at $6.88 and Azure ND96isr_H100_v5 at $12.29 ($8,972 per month). The author notes the gap is not pure margin: Azure’s SKU includes 400 Gbps per GPU of Quantum-2 InfiniBand, while RunPod Community runs on third-party hosts with no data center SLA, and Azure’s single-H100 NC SKU costs $6.98. The comparison covers on-demand list prices only, so reserved and spot pricing would narrow the gap, and AWS and Azure figures are for us-east-1 and eastus. The README gives no collection date, and Lambda is excluded because its listed rates take effect November 16. Source

10. Anonymous GitHub Repo Alleges Gemini Pulls Personal Drive and Gmail Content Into Chats and Routes It to Human Reviewers; Only the Policy Part Is Corroborated

A GitHub repository titled “Documentation of Gemini File Leaking Behavior” drew 513 points on Hacker News on November 9 with claims that Google reversed its 2023 promise that Workspace content is not seen by human reviewers, and that Gemini autonomously retrieves personal Drive files and emails during casual chats without explicit tags. The repo has a README only, three commits and an anonymous-looking account, and offers reproduction steps and one anecdote without screenshots, logs or transcripts; its method relies on asking the model to print what its retrieval tool returned, which does not show that a human viewed any file. Google’s own Gemini help page says a subset of Connected Apps data, including summaries and excerpts of emails and files, is reviewed by human reviewers, and an update dated October 16, 2026 on its 2023 blog post says connected Workspace data can be subject to human review in limited circumstances. The autonomous-ingestion claims remain unverified, and no Google response was found [NEEDS VERIFICATION]. Source