AI Daily · Nov 12, 2026 · Thu
ICREAT AI NEWS
Google releases Gemini Decisions API and Gemini Decision 1 model
Google releases the Gemini Decisions API, powered by the new Gemini Decision 1 model (gemini-decision-1), which can evaluate up to 1,000 yes/no, multiple-choice, and rating questions on text, image, audio, video, and PDF inputs in a single request.
Model releases/updates
Google DeepMindFirst partyPlus 2 sources reportingGoogle DeepMind releases AlphaGenome 2, predicting the effects of DNA variants by cell type
Google DeepMind releases AlphaGenome 2, a conditional DNA sequence model that can predict the effects of DNA variants on gene expression for specific cell types or states.
Product releases/updates
Google Developers BlogFirst partyGoogle releases A2UI 1.0, an open protocol for generative UI
Google releases A2UI 1.0, the first stable version of its open protocol for generative UI that lets AI agents respond with interactive interfaces rather than plain text. Version 1.0 adds bidirectional function calling, accessibility support, and extended metadata, and provides open-source renderer SDKs for Web, Android, and iOS as well as agent SDKs for Python, TypeScript, and Dart.
Databricks BlogFirst partyDatabricks releases a new version of Genie: an AI colleague that autonomously handles complex enterprise work
Databricks releases a new version of Genie, positioned as an autonomous AI colleague that can proactively take on long-running enterprise work, continuously track progress, ask follow-up questions about missing inputs, and act when new information arrives, while users set guardrails and approve key steps.
Databricks BlogFirst partyDatabricks upgrades Genie One into an enterprise AI collaboration workbench
Databricks announces that Genie One becomes the unified entry point for enterprise AI cowork, integrating Genie Search, artifact generation, cross-tool execution, and Unity Gateway governance.
OpenRouter AnnouncementsFirst partyOpenRouter launches beta Extractions API for extracting structured fields from text, images, and PDFs
OpenRouter launches the alpha version of its Extractions API, which lets users describe the fields they need to extract structured values from text, images, and PDFs, initially supporting three models: Cloudflare Clairic, ScaleDown Extract, and Inception Mercury Decide.
Industry news
TechCrunch · AINvidia shares fall as board reportedly prepares to oust CEO Jensen Huang
The Wall Street Journal reports that Nvidia's board could meet as soon as Wednesday to decide the fate of co-founder and CEO Jensen Huang, who has been with the company for 33 years. Huang previously admitted to an extramarital affair with a woman identified by U.S. intelligence as a Chinese spy, and The New York Times reported Wednesday evening that he had another affair with a senior executive who reports directly to him. After the news broke, Nvidia shares fell about 8% in after-hours trading on Wednesday and are down about 11% over the past five days.
The DecoderAnthropic says AI erodes Nvidia CUDA advantage, cutting AMD chip porting from months to three days
In investor materials, Anthropic says it used its in-house frontier model Mythos to port Claude to AMD GPUs in just three days with almost no human intervention, with the model writing nearly 100% of the kernels itself, whereas the traditional engineering approach would take 6 to 12 months.
The DecoderEnterprise customers drive OpenAI quarterly revenue to $10 billion, but it still trails Anthropic
According to The Wall Street Journal and The Information, OpenAI's third-quarter revenue grew about 50% quarter over quarter to $10 billion, while its adjusted operating loss widened slightly from $10.1 billion to $10.9 billion.
Phala confidential AI infrastructure breached, DeepSeek V4 Flash 0731 request data exposed
Phala releases a preliminary incident report saying its infrastructure was breached, and that DeepSeek V4 Flash 0731 requests routed to Phala via OpenRouter may have had their contents exposed between October 21, 2026 and November 8, 2026, 05:35 UTC; affected users need to rotate the API keys, tokens, and passwords used in those requests.
Tips and opinions
Author uses 48 AI agents to analyze NASA lidar data, discovering an unregistered Maya site in Mexico within 72 hours
Using agents such as Gemini, ChatGPT, Claude, and Grok, the author converted about 1,200 km² of NASA lidar data into images, which 48 agents screened one by one, discovering an unregistered Maya site in Mexico that includes pyramids at least 8–12 m tall, plazas, and courtyards.
OpenRouter AnnouncementsFirst partyOpenRouter measures the context cost differences of MCP, CLI, and Skill
OpenRouter compares three integration approaches—MCP, CLI, and Skill—on the same GitHub task: with Claude Sonnet 5.5, CLI used 1,421 input tokens, Skill used 3,315, and the GitHub MCP server used 59,417 when loading its default 46 tools.