AI News November 12, 2026: AlphaGenome 2, Runway Praxis-1, Qwen4-Max and More
Thursday, November 12, 2026, was a heavy day for AI news: Google DeepMind launched a DNA model that reads genetic risk cell by cell, Runway open-sourced a 27-billion-parameter robot brain, Alibaba previewed its next flagship model, ElevenLabs shipped 13 products at once, and Databricks turned its Genie assistant into an always-on agent. At the same time, Nvidia's leadership drama deepened and a Federal Reserve governor warned that AI spending could trigger a market correction.
The short answer: The biggest AI news on November 12, 2026, is DeepMind's AlphaGenome 2 for cell-specific DNA prediction, Runway's open release of the Praxis-1 robot model, Alibaba's Qwen4-Max preview (retiring November 25), ElevenLabs' 13-product launch, and Databricks' always-on Genie agent. Nvidia's board is reportedly preparing to remove CEO Jensen Huang ahead of a November 18 House briefing, and United Airlines is blocking unauthorized AI agents from booking flights.
DeepMind Launches AlphaGenome 2 for Cell-Specific DNA Prediction
Google DeepMind introduced AlphaGenome 2 on November 11, 2026, shifting DNA-sequence prediction from broad tissue averages to the behavior of specific cell types and cell states. The model was trained on more than 107 million profiles and aims to help scientists pinpoint which cells are affected by genetic changes linked to disease — a step toward finding where a disease actually begins.
Access comes through the AlphaGenome API for non-commercial research, and commercial research is restricted to a private preview. DeepMind is explicit that the system is not approved for clinical or diagnostic use. For researchers, the practical change is resolution: instead of asking whether a variant matters in a tissue, they can now ask whether it matters in a particular cell type. Our AI news roundup from November 11 covered the week's earlier model launches, including Anthropic's Claude Fable 5.2.
Runway Open-Sources Praxis-1, a 27B Robot Model
Runway made Praxis-1 generally downloadable on November 11, releasing the weights and fine-tuning software for its video-trained world-action robot model under the Apache 2.0 license. The 27-billion-parameter model reports 46.2% success on RoboLab-120, a robot task benchmark. Open weights mean robotics teams can fine-tune the model on their own hardware and tasks instead of renting access through an API — a meaningful shift for a field where most capable models stay closed.
Alibaba Opens Qwen4-Max Preview — But Only Until November 25
Alibaba Cloud began a time-limited Qwen4-Max-Preview on November 11, giving developers access to a new flagship multimodal model that handles text, image and video understanding, reasoning and tool calls. The catch is access: the preview is Token Plan-only, with no pay-as-you-go option, and Alibaba's documentation says it is scheduled to retire on November 25, 2026. Developers who want to test it should move fast — and should not build anything permanent on a preview that disappears in two weeks.
ElevenLabs Ships 13 Products and Upgrades at Once
AI voice company ElevenLabs announced 13 new products and upgrades on November 12, aimed at making machines interact more naturally with people. The release includes three new voice models shipped over the past month — Scribe v3 for speech-to-text transcription, Eleven v4 for expressive text-to-speech that reads the tone, pacing and context of a script, and Eleven Realtime, a speech-to-speech model for fast voice agents — plus Scribe v3 Medical, a healthcare-optimized transcription model, and Interaction Engine, a system for orchestrating voice models and language models across a conversation. ElevenAgents, the company's platform for voice and chat agents, now covers sales, support, recruitment and operations, and a new Partner Program launched with more than 550 partners in 44 countries.
Databricks Makes Genie an Always-On AI Agent
Databricks launched the second generation of Genie on November 12, positioning it as an always-on autonomous AI agent for enterprise work. The new version runs continuously in the background, handles longer-running tasks without repeated prompts, and is joined by Genie Search — a governed search layer across structured and unstructured data, including Google Drive, Jira and Confluence — plus Ontology Manager for governing business definitions, metrics and reporting lines. Genie and Genie Ontology are generally available; Genie Search, Ontology Manager and the always-on version are rolling out through preview and beta programs. Every action stays auditable through the company's Unity Catalog controls.
Nvidia Board Reportedly Preparing to Remove CEO Jensen Huang
Nvidia's board is reportedly preparing to remove CEO Jensen Huang following allegations involving his personal relationships and an ongoing security investigation, according to reports summarized on November 12. The chairs of three U.S. House committees have requested that Huang attend a closed-door national security briefing on November 18. This follows the board's announcement on November 8 of an independent investigation, after reports that Huang was targeted by Chinese intelligence, and a November 9 report that the Justice Department took over a criminal probe into chip smuggling through Megaspeed. Nvidia shares fell about 5% on November 9. The situation remains fluid, and no board decision has been formally announced.
Fed Governor Warns AI Spending Could Trigger a Market Correction
Federal Reserve Governor Michelle Bowman warned that AI investments could trigger a market correction if financial returns fail to meet expectations, raising concerns about AI-related debt and inflated valuations. The warning lands as Anthropic prepares for one of the largest initial public offerings in history — a reminder that the AI boom is increasingly financed with debt, and that investors are watching whether the revenue catches up with the spending. Our guide to the best AI tools for data analysts explains how teams are actually measuring AI's return on investment.
Quick Hits: United Blocks AI Booking Agents, Gemini Decisions API, ByteDance Assistant
United Airlines is refusing to let unauthorized AI agents book flights, even as OpenAI and Meta build assistants that handle increasingly complex tasks — a dispute over who controls customer data and purchasing decisions when an AI shops on someone's behalf.
Google introduced the Gemini Decisions API for applications that need quick decisions rather than long generated answers. Developers can evaluate up to 1,000 yes/no questions, multiple-choice decisions or scoring criteria in one request, with confidence scores, starting at $0.10 per million input tokens. Google also brought Gemini Live to Apple CarPlay, so iPhone users can hold hands-free AI conversations from their car dashboards.
Perplexity tested whether AI agents could escape secure computing environments, and a Microsoft study found that weight-level changes in fewer than 10 dormant neurons can trigger agent misbehavior while ordinary evaluations appear normal. Separately, NVIDIA is contributing OpenShell, an open-source security runtime for autonomous agents, to the CNCF Sandbox ecosystem.
ByteDance is preparing an AI assistant similar to Meta's Muse, with a possible launch in the coming weeks. And more than 20,000 veterans and transitioning service members have claimed a free year of ChatGPT Plus, according to OpenAI. Data.co also launched a private AI platform and data marketplace with 91 datasets from 25 providers.
FAQ
What is AlphaGenome 2?
AlphaGenome 2 is Google DeepMind's DNA model launched November 11, 2026. It predicts how genetic variations affect regulation in specific cell types, trained on more than 107 million profiles. It is available through a research API for non-commercial use and is not approved for clinical or diagnostic use.
Is Runway Praxis-1 really open source?
Yes. Runway released the Praxis-1 model's weights and fine-tuning software under the Apache 2.0 license on November 11, 2026. The 27-billion-parameter world-action model reports 46.2% success on the RoboLab-120 robot benchmark.
How long is Qwen4-Max-Preview available?
Only until November 25, 2026, according to Alibaba Cloud's documentation. The preview is Token Plan-only, with no pay-as-you-go access, so developers should treat it as a short test window.
What did ElevenLabs launch on November 12, 2026?
Thirteen products and upgrades, including the Scribe v3 transcription model, Eleven v4 text-to-speech, Eleven Realtime speech-to-speech models, Scribe v3 Medical, the Interaction Engine orchestration system, the ElevenAgents platform, and a Partner Program with more than 550 partners in 44 countries.
Is Jensen Huang being removed as Nvidia CEO?
Reports on November 12 said Nvidia's board is preparing to remove Huang, but no formal announcement has been made. House committee chairs have asked him to attend a closed-door national security briefing on November 18. Treat the removal as reported, not confirmed.
Why is United Airlines blocking AI booking agents?
United is refusing unauthorized AI agents the ability to book flights, in a dispute over who controls customer data and purchasing decisions when an AI assistant shops on a traveler's behalf. The fight is part of a wider question about whether airlines or AI companies own the customer relationship.
Related: AI News November 11, 2026 | AI News November 9, 2026 | Best AI Tools for Data Analysts in 2026
Comments
Post a Comment