AI.info

The Pulse — Page 14

Browse The Pulse on AI.info.

• Forever Security Finds Browser Flaws in Five AI Coding Tools

  Embedded browsers in Cursor, GitHub Copilot, OpenAI Codex, VS Code and Claude Desktop broke browser security boundaries, Forever Security says. As of October 29, the researchers said Codex and VS Code flaws remained unpatched.

• MLCommons Wins $3.5 Million to Build an AI Agent Benchmark

  MLCommons will use a $3.5 million, two-year award to develop MLPerf Agents, a benchmark measuring the efficiency of complete AI agent workflows. Its first planned workloads cover software engineering, research and data analysis, and customer service.

• Automattic launches Spacefast for agent-built apps

  Automattic has opened Spacefast, a service that lets users deploy and manage apps through AI agents they already use. The service provides hosting, domains and SQL databases, with built-in app authentication still coming soon.

• EUCLYD Raises More Than €200 Million for AI Inference Systems

  Dutch semiconductor company EUCLYD has closed a Series A round worth more than €200 million to develop energy-efficient AI inference hardware. Samsung, EQT’s Scaleup Europe Fund, Somerset Capital Partners and Innovation Industries co-led the financing, while former ASML chief Peter Wennink joins as chairman.

• Strands Agents Releases a 2B Model for Agent Decisions

  Strands Agents released Decider 2B, an open model designed to make fast yes-or-no and option-selection decisions inside AI agent workflows. The model can run locally, but its developers say it is not suited to complex reasoning, chatbots or coding.

• Rothera Automates Reconciliation for 1.7 Million Daily Transactions

  Rothera has gone live on Smartstream’s Air platform, deploying its Cash and Data modules to automate reconciliation across its CFTC-regulated exchange and clearinghouse. The first two live workflows compare instruments and trades, with additional reconciliation flows planned.

• NVIDIA Researchers Publish a 256K-Context Recipe for Coding Agents

  NVIDIA researchers describe a 256K-context midtraining recipe for preparing Nemotron models for coding-agent tasks. In their tests, 500 billion training tokens raised Nemotron 3.5 Lightning’s SWE-bench Verified score from 60.7% to 64.5% after the same supervised fine-tuning.

• OpenAI’s Agents Leaked 53 ChatGPT User Images

  OpenAI’s agents leaked 53 images from ChatGPT users, with most removed and the rest targeted for takedown. The company has not said when the images were posted or whether they show AI-generated content or real people.

• Tower and NewPhotonics Ship 1.6T Photonics for AI Networks

  Tower Semiconductor and NewPhotonics have begun high-volume shipments of laser-integrated photonic chips supporting 800G and 1.6T optical interconnects. The companies are targeting pluggable and near-packaged optics for scale-out and scale-up AI systems, with 6.4T shipments planned for the first half of 2027.

• General Dynamics Teams With Primordial Labs on Vehicle Voice Control

  General Dynamics Land Systems and Primordial Labs announced an agreement to bring Anura natural-language controls to ground combat vehicles. The software is intended to let crews manage onboard systems without cloud connectivity, but it will not control vehicle weapons.

• Anthropic Pledges $150 Million to Bring Claude to 15+ Agencies

  Anthropic commits $150 million over three years to provide Claude, Claude Code and API credits to Genesis Mission research projects at more than 15 federal agencies. NASA, NIH and the National Science Foundation are among the agencies named; the company also promises training and technical support.

• Flatkey Launches Dedicated NVIDIA B300 Cloud for AI Inference

  Flatkey’s new service gives each customer a dedicated NVIDIA B300 cluster for AI inference, with deployment details set to fit the customer’s workload. The company offers reserved terms of one, three or five years and smaller teams can rent a few GPUs by the hour.

• Confluent Makes Lightning Tables Generally Available on AWS, Adds Query as Fifth Platform Pillar

  Confluent announced new capabilities for building a Streamhouse, an architecture that keeps business context current for AI agents, applications and analytics. Lightning Tables is generally available on AWS, and Query is the fifth pillar of Confluent’s Data Streaming Platform.

• Spain Reports First AI-Agent Personal Data Breach

  Spain’s data protection authority has received its first notification of a personal data breach allegedly carried out by an artificial intelligence agent.

• Pennsylvania Signs AI Ad Disclosure Law, Effective Dec. 21

  Gov. Josh Shapiro signed Pennsylvania’s Artificial Intelligence Deceptive Advertising Act on October 22, 2026. The law requires clear disclosures for certain AI-generated or modified ad content that could create a false impression about goods or services.

• Abacus Releases a 4-Bit Nemotron Fine-Tune for Coding Agents

  Abacus.AI released a four-bit agentic coding fine-tune of NVIDIA’s Nemotron 3.5 Super VL, reporting gains over the NVFP4 base on agentic coding benchmarks. The company published the weights on Hugging Face, while an October 23 tracker listed no independent results yet.

• Vaani Opens Its Voice-Agent Platform to Developers

  Vaani has opened a self-serve platform for building, running and reviewing voice-agent calls. Its pay-as-you-go plan uses prepaid billing and a ₹1-per-minute platform fee, with provider keys available under Enterprise.

• Zetaris launches partner program for governed enterprise AI data

  Zetaris has introduced a partner program offering training, sales resources and joint support for technology partners, systems integrators and service providers. The program is built around the company’s AI Data Harness, which connects AI applications and other workloads to governed data across existing systems.

• Perplexity Releases 0.6B and 9B Multimodal Embedding Models

  Perplexity released two publicly available embedding models that search text, images and rendered PDF pages using token-level vectors. The 0.6B and 9B models share an embedding space, allowing a smaller model to query an index built by the larger one.

• Metaview’s Shared Harness Gets New Agents Running in Under Two Hours, Company Says

  A new Metaview recruiting agent now takes under two hours to stand up, down from several days, the company says. The company describes a shared system built around Cloudflare, Modal and AWS, with agents tested through real sessions and customer requests.

• Meta Puts Idle Inference Capacity to Work on Ad Ranking

  Meta uses temporarily idle AI inference capacity to evaluate more ad candidates and route selected requests to more powerful models, the company says. The company describes safeguards for sudden capacity losses but publishes no numerical ad-performance lift.

• Undo Survey Links AI Coding Tools to Production Outages

  Undo’s 2026 survey found that 81% of engineering leaders reported a production incident or service outage linked to AI coding tools in the previous six months. The survey also found that most respondents saw no faster release cycles, while engineers spent more time debugging than writing code.

• Andon Labs Opens Pion for Autonomous Business Experiments

  Andon Labs is opening Pion, a research-preview platform designed to let persistent AI agents run companies with access to email, banking, phone and browsers. Pion grew from its experiments with vending machines, a retail store, a café and radio stations, the company says.

• Google DeepMind Unveils Genie 4 With Real-Time Audio

  Google DeepMind announced Genie 4 on October 27, 2026, describing a world-simulation model that generates interactive scenes, ambient sound and dialogue. The research preview is limited to trusted testers, while access through Project Genie and developer tools is coming soon.

• AWS Adds SOCI Lazy Loading to SageMaker Training Jobs

  AWS is adding Seekable OCI indexing to Amazon SageMaker AI training and processing jobs, allowing containers to start before their full images download. In AWS benchmarks, a 13.2 GB image started in about six seconds instead of 3 minutes 42 seconds, though the first training epoch took longer as needed files streamed in.

• Australian Inquiry Exposes Microsoft Copilot’s Dangerous Advice

  Australian MP Andrew Hastie told a parliamentary inquiry that Microsoft Copilot suggested replying “congratulations!” to a constituent planning voluntary assisted dying. The hearing also examined Australia’s dependence on US-based AI systems and plans to attract major developers under Australian law.

• Fireworks AI incident affected customer data and incomplete models

  The October security incident affected account and billing information, while a limited set of accounts also had secrets, datasets and incomplete model weights affected. Additional customer data impact was linked, the company says, to an unauthorized workload run through a third-party cloud provider.

• Websites Block AI Agents From Shopping, Booking and Banking

  Users report that AI agents are being blocked or tripped up by security systems on popular websites, while some companies restrict or have not enabled agent access. Cloudflare’s new controls let site owners set separate permissions for search, AI training and agents, with recommended settings for new Cloudflare websites varying by business model.

• Capcom Details REX Plan to Make RE Engine AI-Ready

  Capcom presented REX, a staged upgrade to its RE Engine, at its October 2, 2026, developer conference. The plan focuses on faster development tools and a foundation for AI-assisted coding and testing, not a new engine built from scratch.

• Recursive Releases Flint, a 40B Coding Model for Local Use

  Recursive released Flint, a 40-billion-parameter coding model that activates 2.5 billion parameters per token and is designed to run on local hardware. Its Eureka Machine helped develop the model, the company says, and the weights are available under the Apache 2.0 license.

• Microchip Completes Hailo Acquisition to Expand Edge AI Portfolio

  Microchip Technology has completed its acquisition of Hailo, adding edge AI accelerators, vision processors and robotics technologies to its semiconductor portfolio. The deal brings more than 100 customers and a developer community exceeding 10,000 users, while financial terms remain undisclosed.

• China Central Bank Adviser Says AI May Deepen Weak Demand

  Huang Yiping, a People’s Bank of China monetary policy committee member, warns that AI could widen the gap between China’s strong supply and weak domestic demand. He calls for higher household incomes, market reforms and central borrowing to repair damaged balance sheets.

• Amazon Blocks Meta’s Muse From Shopping on Its Website

  Amazon has blocked Meta’s Muse personal AI agent from shopping on Amazon.com, citing unauthorized access, credential handling and privacy concerns. The dispute adds a new layer to Amazon’s fight over AI agents that browse stores and buy products for customers.

• Oracle Prices $20 Billion Stock and Bond Offering

  Oracle priced a $20 billion stock-and-debt offering on November 2, with the stock sale and five note tranches expected to close November 4. The financing follows Oracle’s October 28 plan to limit additional capital-market funding for its fiscal 2027–2030 investment cycle to $20 billion, separate from a $20 billion at-the-market stock sale completed in the first quarter of fiscal 2027.

• Google Offers Earth AI Data for Public-Health Research

  Google Research presented five public-health studies using its Population Dynamics Foundation Model, which turns aggregated search, mobility and environmental signals into geographic data. The embeddings are available commercially in Preview, while researchers can request no-cost access for select non-operational studies.

• White House Sets $1B-Plus AI Commitments for Genesis Mission

  The White House plans to announce more than $1 billion in industry commitments to President Donald Trump’s Genesis Mission, naming AMD, OpenAI and Anthropic among the participants. Separate summit announcements include a Georgia investment in research computing and more than $100 million for federal AI science tools.

• Affirm Deploys Transformer Model for Real-Time Credit Decisions

  Affirm has deployed a transformer-based underwriting model at U.S. checkout. The system produced 3.4% more completed purchases while maintaining comparable credit risk, the company says.

• Apple Debuts $299 HomeView Smart Display With Siri AI

  Apple introduced HomeView, a 7-inch smart-home display with Siri AI, FaceTime and controls for connected devices. The device starts at $299 in the U.S., with sales beginning November 13, while Apple says it will not be sold in the European Union at launch.

• Google and Constellation Strike 20-Year Deal for 890 MW of Nuclear Power

  Google and Constellation Energy signed a 20-year agreement to add 890 megawatts of nuclear capacity to the PJM grid through upgrades at 11 reactors. A separate 15-year agreement covers another 2,700 megawatts from Constellation’s existing fleet.

• Persona Launches AI Agents for KYC Reviews and Identity Decisions

  Persona launched Persona AI on October 29, a suite that includes agents for identity case reviews and tools for building KYC and KYB workflows. Parafin cut median KYC review time by about 95% using Persona’s Case Review Agents, the company says.