Skip to content
AI Daily / No. 35
Nov 12, 2026 · ThursdayPublished every day at 1:00 AM ET

AI Daily · Nov 12, 2026 · Thursday

BAD INFLUENCE

12stories13sources5company releases2new models
AIWritten from the sources · 4 min read

Google launches Gemini Decisions API for fast multimodal decisions

Google introduces the Gemini Decisions API, powered by the Gemini Decision 1 model, for fast closed-set decisions across text, images, audio, video, and PDFs. The model is priced at $0.10 per 1M input tokens with zero output tokens billed, and it is available through the Gemini API in Google AI Studio.

Google · AI blogOfficial19 more sourcesRead the original
01

Model releases & updates

Google DeepMindOfficial5 more sources

Google DeepMind releases AlphaGenome 2 to predict cell-type-specific DNA variant effects

Google DeepMind announces AlphaGenome 2, a conditional model that predicts how DNA variants affect specific human cell types. It identifies nearly twice as many causal variants as the original model in the TenK10K immune-cell dataset, 30% versus 16%. The model is available today for non-commercial research via the AlphaGenome API and in private preview for commercial research, and it is intended for research use only, not clinical or diagnostic purposes.

LeiphoneNews3 more sources

Baidu releases ERNIE 5.5 Flash Preview with 280B parameters and 1M-token context

Baidu says it released ERNIE 5.5 Flash Preview on November 11, with 280 billion total parameters, 14 billion active per token, and a context window of up to 1 million tokens. The model is available through Baidu Intelligent Cloud's Qianfan platform, with API prices of 2 yuan per million input tokens and 8 yuan per million output tokens during peak hours, discounted 50% for a limited time. Baidu says model weights will be released to the global open-source community on November 25.

02

Product launches & updates

Google Developers BlogOfficial1 more source

Google releases A2UI 1.0 stable protocol for agent-generated interfaces

Google releases A2UI 1.0, the first stable version of its open protocol for generative UI. Agents describe screens, and each app renders them with native components via open-source renderer SDKs for the web, Android, and iOS. Atlassian, Alation, Bloomberg, and ByteDance are already building with it, and the Jetpack Compose renderer currently supports 0.9.1 with 1.0 coming soon.

JetBrains AI BlogOfficial

JetBrains ships Air, an agent workspace inside its IDEs

JetBrains releases Air in its IDEs with the 2026.3 release, letting developers run coding agents such as Junie, Codex, Claude Agent, and GitHub Copilot in one place. Agents can use the IDE's MCP server, inspections, build tools, debugger, and refactorings, and developers review changes with Code Notes. Code completion does not count against JetBrains AI quota, and Junie is free in its Lite configuration. Paid JetBrains AI is not required for local agent workflows, though provider charges and usage limits still apply.

Hacker News · Show HN, AI (20+ points)Blog

Indicode releases Klad, a source-available paint app that coding agents can drive

Indicode BV introduces Klad, a 2D painting and animation app that a coding agent can operate through its command line on the same drawing as a human. The agent's strokes stay editable as data, and a user can recolour or move them later. Klad is source-available rather than open source, free for personal use and for companies with less than 500,000 euros in annual revenue, and it is in an early first public version that users build themselves.

Databricks BlogOfficial2 more sources

Databricks introduces an autonomous Genie AI coworker for ongoing enterprise work

Databricks introduces the new Genie, an autonomous AI coworker that proactively handles long-running enterprise work and takes action as new information arrives. Genie uses Genie Ontology for business context, Genie Search for finding information across sources, and Unity Catalog and Unity Gateway for permissions, approvals, and cost controls. Databricks says that in a benchmark built with Turing, Genie answered correctly in 91% of task runs, 1.6 times the rate of the stronger frontier agent, while using less than half as many tokens.

The DecoderNews1 more source

Grok-built apps exposed patient records and SpaceXSI's sales data

A security firm, Claroc, found that 35,578 of 37,571 still-online apps built with Grok's Build mode were open to anyone with the link. The default "Anyone with the link" setting, plus app addresses that appear in searchable DNS records, exposed patient records, plaintext passwords, and SpaceXSI's own sales documents. SpaceXSI says apps are private until shared and has taken 24 employee apps offline, but has not removed the other open apps.

The DecoderNews3 more sources

Anthropic says AI cut its AMD chip switch from months to three days

Anthropic says its Claude model was running on AMD GPUs three days after work began, with almost no human intervention. The company says Mythos wrote or ported nearly 100 percent of the kernels, compared with six to twelve months for traditional engineering. The figures come from an investor presentation and have not been independently verified.

03

Industry news

TechCrunch · AINews2 more sources

Nvidia board reportedly prepares to oust CEO Jensen Huang

The Wall Street Journal reports that Nvidia's board is preparing to oust CEO Jensen Huang, and Nvidia's shares fell in after-hours trading. The New York Times reported Huang also had an affair with Nvidia vice president Kimberly Powell, which Nvidia, Huang and Powell have not confirmed. The story ties back to U.S. probes into Singapore-based Megaspeed and alleged diversion of Nvidia chips to China.

CNBC · TechnologyNews

Lawmakers call for Nvidia CEO Jensen Huang to resign after China spy affair

Members of Congress, including Senator Kevin Cramer, are calling for Nvidia CEO Jensen Huang to resign after he admitted an affair with a woman U.S. intelligence reportedly concluded was a Chinese spy. Three House chairs gave Huang until November 18 to testify and said subpoenas are possible, while lawmakers link the affair to pending AI chip export control bills in the National Defense Authorization Act.

DeedyX

AMD's 10-Q puts its pending Taalas acquisition at $2.9 billion

AMD's latest 10-Q filing shows the company agreed in August 2026 to acquire Taalas Inc., a developer of specialized inference silicon, for approximately $2.9 billion in a cash and stock transaction. The filing also includes contingent consideration of up to approximately $1.4 billion and says the deal is expected to close by the end of 2026, subject to regulatory approvals.

Image from @deedydas's post
04

In brief

Get AI Daily by email

The day's top AI news, in one email. Unsubscribe anytime.

Every day at 1:00 AM ET

We use your address only for this list. Privacy

End of issue

Bad Influence Daily is automatically edited from public sources. Each item links to its original · Daily archive