AI.info

The Pulse — Page 46

Browse The Pulse on AI.info.

• Apple Code Shows Siri Can Delegate to Claude and ChatGPT

  Private frameworks in iOS 27 and macOS Golden Gate reveal Siri mechanisms for delegating requests to Claude or replacing Apple's server-side model with GPT-5.6. The capabilities are not yet available to ordinary users, but demonstrations show third-party models handling system actions through Siri.

• Eighth Circuit Pauses Minnesota’s AI Nudification Ban in xAI Appeal

  The Eighth Circuit has put Minnesota’s AI nudification ban on hold while xAI appeals a lower court ruling that denied its request to block the law. The appeals court’s order does not resolve xAI’s challenge to the statute under the First Amendment.

• GitHub Opens ReviewBench to Test AI Code-Review Agents

  GitHub launched ReviewBench on October 5, 2026, as an open benchmark for AI code-review agents. Its 219 pull requests, labeled findings and shared scoring process let teams compare reviewers and test their own systems.

• Illumina’s SpliceAI2 Finds 17% More Rare-Disease Variants

  Illumina’s SpliceAI2 model identified 17% more disease-relevant splice variants than other tested splicing models in a rare-disease research dataset, the company says. A preprint reports the comparison used 7,504 probands and found 133 excess panel-matched variants for SpliceAI2 versus 114 for the next-best model at a fixed odds ratio of 2.

• NVIDIA Releases Kumo Tabular With 215M Parameters

  NVIDIA released Kumo Tabular, an open model for classification and regression on structured data, in three sizes from 28 million to 215 million parameters. The model uses labeled rows as context and makes predictions without task-specific training or feature engineering.

• NVIDIA Post Shows Factory Manager Agent Coordinating Specialized Agents on Suspected Defects and Rework

  The NVIDIA Factory Operations Blueprint (FOX) reference example explains how a factory manager agent coordinates specialized agents to inspect suspected defects, arrange rework and investigate likely causes. Its SMT reference example follows an AOI alert through inspection, rework and root cause analysis, while a separate Siemens example covers production planning and inspection preparation.

• Cook Says Apple Intelligence Test Version in China Is Due Later in October

  Apple executive chairman Tim Cook told The Paper in Shanghai that Apple Intelligence has been approved in China and that a test version with real-time translation, Writing Tools and Clean Up arrives later in October, without Siri AI.

• Microsoft Recasts Copilot Around Home, Code and Autopilot

  Microsoft announced a redesigned Copilot experience with Home, Code and Autopilot, alongside a usage-based billing model for agentic work. Home and Code are slated to roll out through the Frontier program, while Autopilot is expanding to private preview.

• GlobalFoundries Expands SiGe Capacity for AI Optical Links

  GlobalFoundries and Marvell expand a multi-year agreement to increase silicon germanium capacity at GF’s Burlington, Vermont facility. The capacity will support optical transceivers, near-packaged optics and co-packaged optics for AI data centers.

• Reservoir Seeks Up to $338 Million in Suno and Anthropic Suits

  Reservoir Media filed separate copyright lawsuits against Suno and Anthropic on October 21, 2026, alleging the companies used its music without permission to train AI models. The complaints seek damages that could total about $337.8 million under alternative statutory-damages calculations and also ask a court to restrict use of the works.

• Gender-Coded Prompts Make AI Writing Less Formal, Johns Hopkins Finds

  Johns Hopkins researchers found that four AI systems produced shorter, less complex and less formal workplace writing when prompts used language associated with women. The study tested GPT-4, Llama, Gemma and Mistral on emails, job applications and resignation letters.

• NVIDIA Invests in Korean Startup Point2 and Plans Investments in Motif Technologies, RLWRLD and Upstage

  NVIDIA announced an investment in Korean AI interconnect startup Point2 and planned investments in Motif Technologies, RLWRLD and Upstage. Korea’s Ministry of Science and ICT and NVIDIA hosted the MSIT-NVIDIA AI Startup Partnership Forum at COEX. Ryu Je-myung joined NVIDIA leaders to meet with over 20 Korean startups.

• Micro-Nano Core Raises More Than 1 Billion Yuan in C2 Round

  Micro-Nano Core’s C2 round raised more than 1 billion yuan, with JianTou Investment leading and three chips already taped out. The company expects to start customer testing in early 2027, while volume supply depends on customer validation and product certification.

• ai& makes Meta’s Muse Spark Open available in Japan

  ai& has made Meta’s Muse Spark Open available through its ai& Inference platform in Japan. The model runs on ai&’s infrastructure in Japan, with usage-based pricing and input and output excluded from model training.

• OpenAI Reveals Habitat Storage System Serving 1 Billion Weekly Users

  OpenAI’s Habitat storage platform now serves products used by more than 1 billion people each week, the company says. The system handles more than 70 million requests per second and stores over 500 petabytes of data across almost 40 regions.

• NVIDIA Expands Cosmos With Training and Evaluation Tools

  NVIDIA expanded Cosmos with data-generation blueprints, model-training tools and simulation checkpoints for physical AI development. The October 21 announcement also cites partner projects involving autonomous-vehicle data, industrial inspection robots and video monitoring.

• NEURA Robotics Unveils Neuron AI for Robots That See, Feel and Remember

  NEURA Robotics introduced Neuron AI, a physical-AI model that combines vision, touch, audio, robot state, language and memory. The company says it trained the model across five robot embodiments and plans an 11-minute laundry-handling demonstration at NVIDIA GTC Berlin from October 20 to 22, 2026.

• NVIDIA Expands Open Models for Robots and Autonomous Cars

  NVIDIA announced new open models and tools for robotics, autonomous vehicles and other fields on October 21. Its physical AI releases include Cosmos 3.5 world models, simulation tools and Alpamayo 2 Edge, a compact model for in-vehicle scene understanding and motion planning.

• Google, NASA Robot Scores 85% in Early Medical-Bag Test

  A NASA collaboration tested a proof-of-concept robotic assistant for future Moon and Mars missions, Google says. In an early laboratory pick-and-place task, the system achieved an 85% baseline success rate while moving a medical bag.

• OpenAI Recasts Its Model Spec Around Five Values

  OpenAI published an updated Model Spec on October 29, organizing intended GPT behavior around five values, including faithfulness to human intent and harm minimization. Its models do not yet fully meet the standard, the company says, and it describes the update as guidance for future training and development.

• Shopify Gives AI Agents Access to More Than 1 Billion Products

  Shopify launched Shop MCP, giving compatible AI agents access to more than 1 billion product listings and tools for search, carts and checkout. Shoppers generally approve purchases through Shop, while an optional setting allows agents to buy within a budget.

• Coyotec’s Los Angeles Audience Jumps 84.9% Among 18-to-34s

  Entravision’s AI co-host Coyotec shares weekday airtime with broadcaster GeeGee Guzman on Los Angeles station José 97.5 FM. Nielsen figures cited by WIRED show the station’s audience of 18-to-34-year-olds rose 84.9% from August to September 2026, as the show expanded to five additional markets.

• General Intuition Raises $220M at a $6.2B Valuation

  The new financing will support General Intuition’s gameplay-trained AI models and expansion of its New York and European teams. The company is opening a waitlist for partners seeking early access, while its investors describe the models’ potential across virtual and physical environments.

• Norway Moves to Restrict AI Glasses in Public Places

  Norway’s government plans to propose a temporary ban on AI glasses in selected public places, including parks, schools and healthcare facilities. The bill’s scope and exceptions remain under review, alongside a separate expert inquiry into lasting rules for wearable technology.

• Frontier Labs’ Researchers Challenge AI’s Race to Deploy

  A 10-day series of resignations, agent breaches and warnings from frontier AI researchers has pushed Anthropic, OpenAI and other labs to reconsider the pace of deployment. The companies now face pressure to prove that outside oversight can keep up with systems they say may soon improve themselves.

• OpenAI Startup Fund Backs Coherence Neuro’s Cancer Implant

  Coherence Neuro announced a $45 million Series A co-led by the OpenAI Startup Fund and Founders Fund for its cancer-monitoring implant. The company is developing a device to record electrical signals associated with tumors and deliver stimulation, but its human testing so far has been brief.

• Google Opens Free AI Video Creation in Vids to Everyone

  Google now lets anyone with a Google Account create AI-generated video in Google Vids at no cost. The update adds Gemini Omni 1.1 Flash, 1080p output, scene controls, SynthID watermarking and expanded tools for Workspace customers.

• Vanta signs agreement to acquire Pillar for AI agent controls

  Vanta has signed a definitive agreement to acquire AI security company Pillar Security. The companies plan to combine Pillar’s runtime protections and testing with Vanta’s trust-management platform, with features scheduled to arrive from January 2027.

• Rocket Puts Policy Gates Between AI Reasoning and Mainframe Execution

  Rocket Software expands its EVA agentic AI platform for mainframe operations and introduces PlanGuard, a policy layer between AI decisions and system execution. Global pilots are testing governed automation across financial services, government, insurance, retail and telecommunications, the company says.

• Meta Brings Confidential Cloud Processing to AI Glasses

  Meta is extending its Private Processing system to AI glasses, using confidential virtual machines to process personal context in the cloud. The company says the design prevents Meta, host operating systems and infrastructure operators from accessing data while it is being processed.

• ASML Says Planned Tool Supply Will Fall Short Through 2028

  ASML says its planned chipmaking-tool supply will not meet customer demand in 2027 or 2028 as AI investment lifts equipment forecasts. The company is exploring additional capacity for 2028 and raised its 2026 sales outlook to €44 billion to €45 billion.

• Shield AI and Navy Commit $300M More to X-BAT

  Announced public and private commitments to Shield AI’s X-BAT aircraft have reached $400 million after a new $150 million Navy agreement modification. The company will match that funding, and flight testing is scheduled for later in 2026.

• Mistral Acquires SkipLabs as Its Team Forms a New Language Team

  Mistral has acquired SkipLabs and says its team will form a new Language team. The group will work on tools to help coding models write and verify software.

• Qualcomm and OKSI Put Autonomy Software on the QCS8550

  Qualcomm Government Technologies and OKSI announced a collaboration to run OKSI’s OMNISCIENCE autonomy software on Qualcomm processors. The software is optimized for the QCS8550 and is set to be demonstrated at the AUSA Annual Meeting in Washington, D.C., on October 12–14, 2026.

• Tempus Targets 100,000 Whole Genomes for AI Research

  Tempus AI is building a de-identified dataset of 100,000 disease-specific whole genomes linked to longitudinal clinical outcomes. The company plans to make the platform broadly available by mid-2027 and eventually expand it to one million genomes.

• India Launches DEPA Handbook for Trusted AI Training

  India’s Ministry of Electronics and Information Technology launched a handbook for using its Data Empowerment and Protection Architecture in AI training. The framework combines contractual permissions, controlled computing environments and data provenance so institutions can collaborate while retaining control of their data.

• LangChain Adds Browser Control and Async Subagents to v0.10

  Managed Deep Agents v0.10 adds browser tools for websites without APIs and lets subagents handle tasks in the background while the main agent continues talking with users. The public beta also changes Slack run controls and removes some project configuration options.

• Restate Raises $20M to Make AI Agents Survive Failures

  Restate raised a $20 million Series A to expand its durable-execution platform for AI agents and backend workflows. The Berlin-based startup plans to hire engineers and go-to-market staff as it takes on Temporal.

• Infinigence AI Raises RMB 2.3 Billion for Chip-Agnostic Infrastructure

  China’s Infinigence AI has completed a RMB 2.3 billion Series C, bringing its cumulative funding to RMB 4.3 billion. The company says its platform connects computing capacity across 16 chip types, while its announcement does not identify investors in this round.

• MIT Tests Suicide-Risk Model on 16,000 Crisis Chats

  MIT researchers tested a lightweight text model on about 16,000 crisis-counseling conversations, using a lexicon linked to 49 suicide-risk factors. The tool can flag patterns in text, but the researchers say it needs thorough validation and human oversight before clinical use.