AI News, 17 November 2026

Tuesday 17 November

The AI data center boom is sapping the strength of the Top500

Supercomputer concept image showing server room with open racks and multi-colored data flows running across room.

AI-generated summary The latest Top500 list reports that the highest performing chips are being diverted to AI hyperscalers. Epoch AI reported in February 2026 that the aggregate computing power of AI chips sold each quarter was growing 2.5x a year.

Earlier:

London’s Model ML raises $250M from General Atlantic and QED to put AI agents inside banks — TFN

AI-generated summary Model ML secured $250 million in Series B funding led by General Atlantic and QED Investors to deploy AI agents inside banks. The London-based company reached a $1.5 billion valuation and reported that customer data is never used to train its AI models.

Also covered byPulse 2.0HuggingNewsTech.eu

Bland Introduces Tofu-1: A Full-Duplex Speech-to-Speech AI Model for Challenging Call Environments

Bland Introduces Tofu-1: A Full-Duplex Speech-to-Speech AI Model for Challenging Call Environments

AI-generated summary Voice-AI company Bland released Tofu-1 in research preview on November 15, 2026, as a full-duplex speech-to-speech model for phone calls. Bland states that Tofu-1 was trained on millions of hours of real conversation data and features closed weights managed exclusively by Bland.

Earlier:

Adobe MAX 2026 Expands Firefly into an AI Creative Studio: Breakdown of GA, Beta, and Private Beta Features

Figure 1 Adobe official documentation diagram illustrating Firefly Graph node workflows

AI-generated summary During Adobe MAX 2026 on November 10, 2026, Adobe announced an expansion of Adobe Firefly into an AI creative studio with features organized across general availability, beta, and private beta tiers. Adobe reports that Premiere mobile on iPhone adds licensed music from UMG and WMG for personal videos while commercial projects rely on Adobe Stock Audio.

Earlier:

Tencent releases Hy4 flash with 12 billion active parameters and lower task costs

Tencent headquarters towers under construction beside a prominent bus stop sign in Shenzhen

AI-generated summary Tencent released Hy4 flash on 16 November 2026 with 12 billion active parameters and a 1-million-token context window for routine workplace tasks. Tencent says the model is available on platforms including Yuanbao and WorkBuddy, with free access for WorkBuddy continuing through 30 November.

Also covered byFresh WeightsRSS.bze/acc
Earlier:

Jensen Huang was building Anthropic's worst nightmare. Then he was ousted.

AI-generated summary Nvidia was investing heavily in cheaper and more open AI models that competed with offerings from Anthropic. Anthropic is preparing for a stock market debut this week that could value the company at $1.4 trillion.

Also covered byZeniteq
Earlier:

What the therapist bot heard: Character.ai's chats are on the block

AI-generated summary The Consumer Federation of America wrote to the judge in Character.ai's bankruptcy regarding the sensitivity of user chat logs and the risk of abuse by data brokers, advertising technology companies, and surveillance contractors. The letter stated that users' chats are built into the company's language models and noted precedent for demanding algorithmic disgorgement.

Also covered byFast Company

Filmustage vs ScreenWeaver: AI Script Breakdown or Script to Film

Filmustage's name and the filmustage.com domain next to the ScreenWeaver storyboard editor, on a dark background with a violet glow

AI-generated summary Filmustage is an artificial intelligence pre-production platform that analyzes screenplays to build schedules, budgets, and call sheets for live-action shoots. ScreenWeaver begins with a script to generate storyboards and turn every shot into artificial intelligence video using selected models.

Earlier: