AI.info

The Pulse — Page 21

Browse The Pulse on AI.info.

• Google Puts Gemini 4 Argon Into Robotics Workflows

  Google presents Gemini 4 Argon in robotics examples spanning robot control, simulation and physical-task analysis. The demonstrations report specific results, but rely on defined robot setups, simulation tools and task-specific workflows.

• Atlassian Launches AMP for Human-Agent Teamwork

  Atlassian announced the Agentic Multiplayer Protocol, or AMP, at Team ’26 Europe in Amsterdam on October 6, 2026. The framework connects agents to Atlassian’s Teamwork Graph and adds shared work features, review steps and controls for agent access.

• Judge Sentences Michael Smith to 18 Months for AI Streaming Fraud

  U.S. District Judge John G. Koeltl sentenced Michael Smith to 18 months in prison for using bot accounts to generate fraudulent music streams and collect royalties. Smith also received two years of supervised release and was ordered to forfeit $8,091,843.64.

• ACLU Warns AI License-Plate Analysis Is Driving Police Stops

  The ACLU says police and federal agents are using license-plate data and AI tools to flag drivers based on where they travel. A Border Patrol investigation by the Associated Press found that officials can treat ordinary routes and trips as suspicious, while CBP says its system serves defined security purposes.

• METR Seeks Access to Frontier Labs for AI Risk Assessments

  METR proposes independent assessments of how frontier AI labs manage misaligned agents, monitor internal systems and track AI-driven research acceleration. The voluntary process would let labs limit assessor access and redact reports, which METR says cannot establish that development is safe.

• JetBrains Releases Mellum2.1 for Coding Agents

  JetBrains has released Mellum2.1, an Apache 2.0 model designed to work in code repositories and support coding agents. In JetBrains’ tests, it resolved 47% of SWE-bench Verified tasks, compared with 2% for Mellum2 and 50% for Qwen3.5-9B.

• Ecosia Says Its Mistral Exit Is About Open Models, Not China

  Ecosia founder Christian Kroll says the search company is shifting from Mistral toward open AI models accessed through Germany’s Melious platform. He says the priority is cost, flexibility and energy use—not where a model was developed.

• Microsoft and Coinbase Take Down AI Phishing Service EvilTokens

  A coordinated operation disrupted EvilTokens, a phishing-as-a-service platform that used AI to analyze compromised inboxes and target cryptocurrency payments, Coinbase says.

• Ping An Bank Board Approves First AI Management Rules

  Ping An Bank’s board approved internal rules for artificial intelligence on September 29, making it the first listed mainland Chinese bank to publicly disclose such a policy. The rules’ full text remains unpublished as regulators press banks and insurers to formalize AI oversight.

• HHS Drops AI Model-Card Rules From Health IT Certification

  HHS finalized changes removing federal certification requirements for disclosures about certain clinical decision-support tools, despite commenters’ concerns about transparency and provider oversight. The rule removes specific disclosure and risk-management duties while encouraging health care organizations to ask developers about tools’ quality and performance.

• Modal Raises $750 Million at a $15.75 Billion Valuation

  Modal announced a $750 million Series D led by Accel on October 14, valuing the AI infrastructure company at $15.75 billion. Its annualized run rate has passed $1 billion, while its Sandboxes product has reached a $300 million run rate, the company says.

• Dream Labs Raises $90M and Opens Its Dream-1 Robot API

  Dream-1 is available through a public API and, Dream Labs says, can control unfamiliar bimanual robots with parallel grippers. The company also shows fine-tuned examples on robots with dexterous hands and other end effectors, after collecting robot-specific data.

• MIND Raises $72M to Automate Data Loss Prevention for AI

  Seattle-based cybersecurity company MIND has raised $72 million in Series B funding led by Crosspoint Capital Partners. The company will use the capital to expand its AI-native data loss prevention platform, partnerships and enterprise operations.

• OpenAI Gives ChatGPT Control of Chrome Tabs

  OpenAI updated its ChatGPT Chrome extension on October 15, 2026, adding the ability to read across open tabs and carry out browser tasks. The extension can fill forms, send messages, download files and organize tabs, with website-access controls managed through ChatGPT.

• Profound raises $180M at $1.8B valuation for AI-search marketing

  Profound has raised a $180 million Series D at a $1.8 billion valuation to help brands appear in AI-generated search answers. Sequoia Capital and Kleiner Perkins led the round, which comes less than seven months after the company’s $96 million Series C.

• Anthropic’s Claude Watermark Leaves a Sampling Loophole

  Anthropic plans to add text watermarks to four older Claude models, but says API users should keep default sampling settings on Opus 4.6 and Sonnet 4.6 if they rely on the mark. The caveat gives developers a way to avoid watermarking those models by changing a sampling parameter.

• AGIBOT Pairs GO-3 Preview With a New Measure of Robot Work

  AGIBOT introduced Worken, a proposed measure of productive robot work, alongside its GO-3 Preview model at an event in Shanghai. The framework combines active robots, autonomous working hours and task-specific output efficiency, the company says.

• Google Keeps AI Overviews in School Search as Districts Push Back

  Google will not give schools an option to disable AI Overviews while keeping traditional Search, leaving districts to consider workarounds or alternatives. Educators and parents say the feature could make it harder for students to assess sources and learn without AI-generated answers.

• Finland Opens Criminal Probe Into Google’s Muhos Data-Center Work

  Finnish police have opened a criminal investigation into preparatory work at Google subsidiary Tuike Finland’s planned data center in Muhos. Investigators cite preliminary concerns about drainage and water management; the case is expected to continue into 2027.

• Bessent: U.S. proposes AI incident alert system in talks with China

  Treasury Secretary Scott Bessent said the United States proposed a notification mechanism for artificial intelligence incidents that could affect national security during weekend discussions with Chinese officials.

• Socure Reports 99.99% Lab Accuracy Identifying Muse and Instinct

  Socure’s Agent Detection system identified the consumer AI agents Muse and Instinct with 99.99% accuracy across 700 behavioral variables in controlled testing, the company says. The company has made detection generally available, opened Agent TrustScore to beta customers and scheduled Agent Credential for general availability in the first quarter of 2027.

• Goldman Sachs Raises Data-Center Power Forecast on AI Token Growth

  Goldman Sachs now expects global data-center power demand to rise 170% by 2030 from 2025 levels, up from a 117% estimate made in March. Analysts cite higher hyperscaler spending, more AI-server shipments and expanding token consumption despite efficiency gains.

• SK hynix Launches Silicon Valley Venture Arm for AI Infrastructure

  SK hynix has launched SK hynix Ventures, a corporate venture capital brand focused on AI computing, data centers, system software and optical interconnects. The Korean memory-chip company introduced the venture arm at its first Ventures Day in Silicon Valley on September 17, 2026.

• CMU Researchers Recast Foundation-Model Training as One Algorithm

  A study proposes using one likelihood-based training rule across several foundation-model training stages, with data mixtures changing over time. Experiments found tradeoffs among reasoning performance, retention and data schedules, rather than one mixture that wins across settings.

• WhiteMatter raises $20 million for enterprise AI agent controls

  WhiteMatter raised $20 million in a Series A led by Blumberg Capital. The company says it has 2.25 million agents under management and will use the funding for research, development and expansion.

• SoftBank’s Masayoshi Son Warns AI Could Become ‘Super Dangerous’

  SoftBank Group CEO Masayoshi Son warned in Kyoto on October 4 that powerful AI could become dangerous in the wrong hands. His remarks came alongside a 17-country initiative to expand AI’s role in scientific research.

• vLLM Adds Push and Reverse KV Transfer for Agent Workloads

  vLLM’s bidirectional KV transfer cut time to first token by up to 3.3 times in a long-context GLM-5-FP8 test, while a separate push connector improved TTFT by up to 3 times for Qwen3-32B. The two approaches target different inference workloads, and vLLM’s guidance favors pull-based transfer when agents need to reuse KV across conversation turns.

• Waymo Starts Toronto Mapping Ahead of 2028 Robotaxi Launch

  Waymo will begin manually driven mapping and testing on Toronto streets as it prepares a fully autonomous ride-hailing service. The company is targeting 2028 for its first riders and plans to work with Element Fleet Management on local operations.

• Meta Grants $2 Million to Five Universities for AI Research

  Meta is providing $2 million and access to its AI models to five U.S. universities for research in medicine and other sciences. Projects include monitoring infant tracheostomy tubes, studying dementia conversations and assessing heart-failure risk.

• AWS Brings Bedrock Models to Salesforce as Agents Move Into Slack

  AWS and Salesforce announced new integrations connecting Amazon Bedrock models, Salesforce Agentforce, Amazon Quick and Slack. The September 15 agreement also adds zero-copy data access, AWS agents in Slack and planned voice coordination between Agentforce and Amazon Connect Customer.

• Governments and AI companies sign Paris safety declaration

  A declaration signed by governments, AI companies and civil society groups sets out three commitments on frontier AI safety. Shared evaluation methods, stronger accountability and better systems for reporting incidents and emerging risks are called for in the declaration.

• Lambda Plans to Build a Data Center Campus in Lowndes County, Alabama, With an Estimated $33.1 Billion Investment

  Lambda plans a proposed data center campus in Lowndes County, Alabama, with an estimated $33.1 billion investment. Lambda says it will pay the full cost of the facility’s power and grid infrastructure, and construction is expected to begin in the first quarter of 2027.

• Australia’s Digital Duty of Care Bill Brings AI Chatbots Under Safety Rules

  Australia introduced its Digital Duty of Care Bill on October 15, 2026, setting proposed online safety requirements for services including AI chatbots. The bill also covers social media feeds, child safety, complaint handling and penalties of up to A$109.2 million.

• Parallel Systems Raises $100 Million to Scale Autonomous Railcars

  Parallel Systems has raised a $100 million Series C led by AVP to expand production of its autonomous, battery-powered Panther freight vehicles. The company is testing the vehicles on 160 miles of track near the Port of Savannah as it works toward its first commercial freight movement.

• Comer Seeks Records on Delaware’s AI-Run Company Plan

  House Oversight Chairman James Comer has asked Delaware Gov. Matt Meyer for records on a proposal to let AI agents manage companies. The request raises questions about liability protections, safety reviews and the role of private-sector partners in drafting the plan.

• ElevenLabs Reaches $22 Billion Valuation in $300 Million Tender

  A $300 million employee tender offer values voice AI company ElevenLabs at $22 billion, double its Series D valuation in February. Enterprise customers now account for 55% of revenue and its agents handle more than 15 million conversations a week, the company says.

• Nscale Lines Up $3.36B in Convertible Funding Ahead of IPO

  Nscale announced $3.36 billion in convertible loan-note financing, including $2.36 billion at closing and a further $1 billion commitment from Nvidia expected in November. The capital will support expansion of its AI cloud platform, the company said, while the notes are set to convert into shares when its IPO is completed.

• India’s NMC Bans Commercial AI Medical Campaigns

  India’s National Medical Commission has banned AI-generated promotional campaigns that advance commercial interests in healthcare. Its new guidelines also require an AI source mark on compliant AI-generated promotional content and set penalties for violations.

• IBM Puts Chat, Decisions and Speech in Granite Switch 4.1

  IBM’s Granite Switch 4.1 combines chat, decision and safety adapters with an embedded speech-recognition encoder in 8B and 30B checkpoints. The speech path transcribes English audio into text for the language model, with lowercase, unpunctuated output and no joint training of the encoder and language model.

• Jack & Jill raises $40M for two-sided recruiting agents

  Jack & Jill has raised $40 million in Series A funding to build a recruiting service in which an AI agent represents the candidate and another represents the employer.