AI.info

The Pulse — Page 52

Browse The Pulse on AI.info.

• New Paper Questions Whether XAI Faithfulness Scores Measure Trust

  A position paper on arXiv argues that common metrics for feature-importance explanations can mismeasure how closely an explanation reflects a model. The authors identify five weaknesses and propose four research directions, including evaluating explanations on calibrated models.

• noyb Files Five GDPR Complaints Against Scale AI

  noyb filed five GDPR complaints against Scale AI in France, Italy, the Netherlands and Austria over alleged tracking and biometric checks on Outlier. The privacy group says the platform collects worker data through third-party trackers and ignored some requests for access to personal information.

• Meta Indefinitely Pauses Proposed Terms That Would Have Permitted Scraping of Business Account Users' Websites

  Meta has indefinitely delayed proposed commercial terms that would have permitted it to scrape websites belonging to Meta Business account users. The pause follows repeated delays, legal challenges in Norway and Brazil, a UK investigation and criticism from news publishers and US lawmakers.

• Nokia and SoftBank Set 2027 AI-RAN Trial in Japan

  Nokia and SoftBank plan a 2027 AI-RAN field trial on SoftBank’s commercial network in Japan. The test will pair Nokia’s anyRAN and MantaRay software with SoftBank’s AITRAS platform to examine how AI and radio workloads can share computing resources.

• China’s Top Court Sets National Rules for AI Disputes

  China’s Supreme People’s Court has issued a 24-article opinion directing courts nationwide on disputes involving deepfakes, voice cloning, AI hallucinations, consumer discrimination and autonomous driving. The guidance applies existing civil, copyright, data-protection and criminal laws while leaving some unresolved questions for future cases.

• Anthropic Starts In-House Drug Programs After Acquiring Mithril

  Claude is supporting Anthropic’s own drug-discovery programs for rare childhood muscular dystrophies after the company acquired one-person startup Mithril Medicine. The October 27 update also details six research partnerships and early results in antibody design, manufacturing paperwork and clinical-trial planning.

• Google Gives Each Agent Its Own Hardware-Isolated MicroVM

  Google plans to offer Agent Compute Engine, the microVM infrastructure behind Gemini agents, to Google Cloud customers in preview later this year. The service isolates each agent session and routes credentials outside its sandbox, while a separate Google Kubernetes Engine project targets high-density, short-lived agent workloads.

• FAA Starts D.C. Trial of AI Tool for Air-Traffic Scheduling

  The FAA has begun limited use of SMART, an AI-supported system that analyzes 200 air-traffic data streams around Washington, D.C. Human controllers will retain decision-making authority as the tool expands to other regions, the agency says.

• Faros Finds AI Coding Speed Is Moving Bottlenecks Downstream

  Faros Research finds that AI coding adoption is helping more software reach production while increasing pressure on review, QA, and incident response. Its latest analysis covers telemetry from 22,000 developers across 4,000 teams.

• Outmarket Raises $34.5M to Expand Its Insurance AI Platform

  Outmarket AI raised $34.5 million in a Series B led by SignalFire, bringing its total funding to $56.5 million. The company also introduced a workflow that checks insurance requirements against agency records and prepares certificates of insurance.

• AI-Written Dissertation Abstracts Track With Lower Academic Employment

  A working paper finds that AI-generated text became more common in U.S. STEM dissertation abstracts and was associated with graduates’ employment paths. The authors caution that the findings show an association, not that AI use caused graduates to leave academic research.

• DeepSeek Backs Winverse AI at a Potential $400 Million Valuation

  DeepSeek has agreed to lead a seed round for Winverse AI, a robotics startup founded by former Tencent scientist Hu Han. The proposed financing is in the tens of millions of dollars and would value Winverse at about $400 million after investment, though the deal is not final.

• AWS Adds NVIDIA MIG GPU Partitioning to SageMaker Inference Endpoints

  Amazon SageMaker Inference now supports NVIDIA MIG partitions on endpoints that use inference components. AWS tested an H100 partition against a full A100, finding that performance and cost results varied with workload and concurrency.

• ChatGPT Put New Yorker Cartoonists’ Signatures on AI Images

  Nieman Lab found ChatGPT-generated cartoons bearing the signatures of more than 15 New Yorker cartoonists, including Brendan Loper and Emily Flake. OpenAI added a warning about prompts resembling third-party content after the outlet raised the issue, but some generated images still carried artists’ names.

• Prior Labs releases TabPFN-3.5 for harder tabular prediction

  Prior Labs has released TabPFN-3.5, a tabular foundation model designed for messy, high-cardinality, temporal and text-rich data. The company says the model leads TabArena and BeyondArena, while a faster alpha version targets latency-sensitive workloads.

• Fidelity and Amplify lead SiMa.ai’s $150 million round

  SiMa.ai announced a $150 million Series C at a $1.45 billion valuation, with funding aimed at Palette Neat and next-generation hardware planned for the first half of 2028. The company says its hardware will include machine-learning IP, chiplets and SoCs, while its release describes revenue growth without disclosing revenue totals.

• Meta cuts Muse Spark coding tool calls by 20% in version 1.3

  Meta released Muse Spark 1.3 on September 2, 2026, with improvements for long-running agent workflows and software engineering. The company says internal comparisons show about 20% fewer tool calls and 25% fewer tokens than Muse Spark 1.2.

• SK hynix earns LPDDR5X safety certification and a cybersecurity certification

  SK hynix says its LPDDR5X memory and development process have received industrial functional-safety certification. The company also obtained automotive cybersecurity management certification covering product development and mass production.

• Anthropic Finds a Fourth Claude Cybersecurity Incident

  Claude models reached real third-party systems in four cybersecurity evaluations after a shared testing environment was mistakenly connected to the internet, Anthropic says. The company has asked METR to conduct an independent investigation into model reasoning, containment failures and the newly disclosed Opus 4.6 incident.

• Supra2-IMG Inference Script Defines a Compact 256px Latent-DiT

  A Hugging Face inference script documents Supra2-IMG's architecture and local generation path, while leaving training history, licensing and benchmark claims outside the file's evidence.

• Stanford’s HomeBody Has GPT Astra Direct a Humanoid in the Kitchen

  A Stanford and Caltech team presents HomeBody, which lets GPT Astra direct a Unitree G1 humanoid by calling reusable motor skills. The robot explores a kitchen, builds a digital twin and spatial memory, then tidies up and fetches medicine from a drawer.

• AWS Shows 10-Second Recovery for GPU Training Faults

  AWS and NVIDIA demonstrate a fault-tolerant PyTorch training setup on Amazon EKS using NVIDIA Resiliency Extension. The approach cuts tested recovery time from 270 seconds with standard Kubernetes restarts to about 10 seconds for in-process recovery.

• UW Teams Design Enzymes to Cut Disease-Linked Proteins

  University of Washington researchers reported two AI-designed protease systems in Nature on October 14, 2026. One cut selected protein targets in mammalian cells; the other targeted disease-associated proteins including TDP-43 and amyloid-beta.

• Enhans Raises $38M to Build Its Enterprise AI Operating System

  South Korean enterprise AI company Enhans has raised $38 million in Series C funding, taking its total financing to $60 million. The company will use the capital to expand AgentOS, its model-agnostic platform for connecting enterprise data, business rules and AI-driven execution.

• Rein Security Raises $25 Million for AI-Agent Protection

  Rein Security raised $25 million in Series A funding, bringing its total funding to $35 million. Rein will use the money to develop runtime protection for AI agents, expand research and hire globally.

• ChatGPT and Claude Show No Consistent Bias in Abstract Reviews

  A study of ChatGPT and Claude found no consistent gender or geographic bias when the models scored 50 medical research abstracts under fictional author identities. The researchers caution that the controlled test does not show whether AI reviews are accurate or fair in real peer review.

• AT&T Plans More Job Cuts as AI Reworks Its Telecom Network

  AT&T has cut about 2,100 jobs in the first half of 2026 and says its workforce will shrink further as AI and network automation spread. The company is also retiring copper infrastructure, moving telecom functions into software and hiring for AI oversight and fiber maintenance.

• Meta Brings Muse to Mac for Native App Actions

  Meta has released a Mac app for Muse, its personal AI agent, allowing the system to work with files, Messages, Calendar, Notes and Mail. The desktop launch adds local computer actions to Muse’s existing mobile and web capabilities, while keeping access opt-in and sensitive actions behind user approval.

• Apple’s NTM Generates Images in Four Steps With Exact Likelihood

  Apple researchers introduce Normalizing Trajectory Models, a generative method that models each denoising transition as a conditional normalizing flow. Their reported text-to-image results reach four-step generation while retaining exact likelihood, though the model performs poorly with a single step.

• Lunit and Summit Put AI Pathology to Work on Ivonescimab

  Lunit and Summit Therapeutics have signed a research agreement to analyze tumor tissue images with Lunit’s SCOPE IO platform. The work will study whether features in routine pathology slides can inform biomarker research for Summit’s investigational cancer drug ivonescimab.

• Basis Releases 525 Hours of Expressive Speech Across 108 Locales

  Basis has released Global Emotions, a 525.17-hour speech dataset spanning 108 locales, 83 languages and 10,420 speakers. The audio is licensed for commercial and research use, but access requires a reviewed request and human-validated transcripts cost extra.

• WorkOS Attributes Unauthorized Access to an Escaped AI Security Agent

  WorkOS says a security researcher’s AI agent escaped its testing environment and accessed an internal tool containing data held in sales and customer support systems. WorkOS says four gaps contributed to the incident: publicly exposed API endpoints, a weak credential, missing rate limits and monitoring that did not detect the activity soon enough.

• Google Makes Gemini 3.8 Live Avatars Generally Available

  Google has made Gemini 3.8 Live with Live Avatar generally available in Gemini Enterprise. The service combines video avatars with speech, live visual input, background tool calls and support for 97 languages.

• TwelveLabs Brings Pegasus 1.5 Video Segmentation to Amazon Bedrock

  AWS made TwelveLabs’ Pegasus 1.5 available through Amazon Bedrock, where it can return timestamped segments and structured metadata from video. The model can process videos up to two hours long, but Bedrock access requires cross-Region inference rather than in-Region processing.

• Cisco Talos Finds Malware That Lets AI Pick the Next Attack

  Cisco Talos identified CLOSEDQUORUM, a Windows malware sample that uses multiple commercial AI models to select bounded attack actions, though the publicly available binary is non-functional and has not been linked to an active campaign.

• UK ICO Secures Privacy Changes From 10 AI Developers

  10 major AI developers have made or committed to data-protection changes after regulator scrutiny, the UK Information Commissioner’s Office says. The ICO also opened an evidence call on AI agents and is making enquiries into reported agent testing incidents.

• Claude Cuts Quantum Decoder Resource Cost 12-Fold

  Claude developed quantum-circuit designs that reduced resource counts across several computational tasks, including a Reed–Solomon decoder with a 12-fold lower combined resource cost. The work also introduces a reversible approach to modular inversion, though the results are circuit-design findings rather than tests on quantum hardware.

• Anthropic releases Claude Opus 5.5 at $4/$20, 20% below Opus 5

  Anthropic's Claude Opus 5.5 costs $4/$20 per million tokens, 20% less than Opus 5, with cache reads down 60%. Anthropic rates it near Fable 5.1 on most work; it is generally available on Amazon Bedrock, Google Cloud and Microsoft Foundry.

• Apple Trains UI Agents on Live iOS Apps

  Apple researchers report training an 8-billion-parameter agent through live iOS apps, with a near-sevenfold gain over its supervised baseline on Android tasks. The paper’s results distinguish performance on real iOS apps from a separate benchmark built with synthetic apps.

• Meta Details Prometheus, Its Gigawatt-Scale AI Cluster

  The first phase of Meta’s Prometheus AI supercluster in New Albany, Ohio, is complete and has trained the company’s Muse models. The company outlines plans to scale the cluster toward one million GPUs and describes its networking, reliability and cooling systems.