AI.info

The Pulse — Page 51

Browse The Pulse on AI.info.

• Altman Makes OpenAI’s IPO Wait on Safety Claims

  Sam Altman said OpenAI needs stronger safety assurances before pursuing an IPO, but warned that waiting too long to go public could also be harmful. At DevDay on September 29, he gave no timeline and said safety and alignment must advance ahead of model capabilities.

• OpenAI Contractors Removed After Using AI to Grade ChatGPT

  Multiple contractors working on OpenAI model-evaluation projects were fired or removed after using AI tools to complete assignments, according to 404 Media. Mercor, one staffing company involved, says its contracts prohibit large language models from performing the work.

• SK hynix Launches Ventures Arm for AI Infrastructure

  SK hynix has launched SK hynix Ventures, expanding its corporate venture capital focus from semiconductor startups to technologies across the AI ecosystem.

• Claude Opus 5.5 (xhigh) Scores 18.5% on ProgramBench With mini-SWE-agent

  Claude Opus 5.5 (xhigh), evaluated with mini-SWE-agent, fully solves 18.5% of ProgramBench’s 200 software-reconstruction tasks and passes at least 95% of behavioral tests on 70.5% of them. The latter figure includes fully solved tasks, and the benchmark paper describes how the evaluation tests agents’ ability to recreate software behavior.

• Turing Institute Starts £2M Programme on Transformative AI Security

  The Alan Turing Institute has secured a £2 million grant for research into the security risks of transformative AI, with work focused on government preparedness, national security and resilience.

• Microsoft Offers Voluntary Exits to 60,000 Workers

  Microsoft is offering eligible employees in participating organizations in four countries a voluntary exit package, with potential reach of about 60,000 workers. The program is time-limited and offers financial support and career-planning resources, but the company has not disclosed a payout formula or expected number of departures.

• Russia's Election Chief Says AI-Linked Groups Launched 1,000 DDoS Waves

  Russia's electoral chief said more than 1,000 waves of cyberattacks targeted infrastructure involved in the country's parliamentary vote, while the defence ministry accused Ukraine of trying to disrupt the election.

• Pentagon Weighs $5B Fluidstack Loan for AI Power and Cooling Gear

  The Pentagon’s Office of Strategic Capital is in talks to lend roughly $5 billion to AI cloud startup Fluidstack to expand U.S. manufacturing of data-center power and cooling equipment. The deal, advised by Palmer Luckey’s Erebor Bank, would be the office’s largest loan and remains unfinalized.

• Abu Dhabi Puts AI Rule Checks Into Production With OpenAI

  Abu Dhabi says its first Machine-Readable Government pilot is already in production, helping officials assess food-import compliance against selected rules. OpenAI models process the checks on GPUs in UAE data centres, while government officials retain responsibility for final decisions.

• webAI Releases Mac App for On-Device AI Without Per-Token Charges

  webAI has released a Mac app that runs its small reasoning models on a user’s computer, which the company says lets people work with private files without sending them to the cloud or paying per token. The app uses user-created Personas that can work together and connect with Slack, DocuSign, Hugging Face and Apple apps.

• Dartmouth Provost Santiago Schnell Resigns After AI Review

  Dartmouth Provost Santiago Schnell resigned on October 28, 2026, after a faculty-led committee found undisclosed AI use in his writing. The committee identified publication-policy violations in a Washington Post column and an academic paper, while saying it could not conclude that a third paper violated its publisher’s policy.

• Satlyt Raises $8 Million to Run AI on Existing Satellites

  Satlyt raised an $8 million seed round led by Non Sibi Ventures to expand software that runs AI and processes data aboard satellites. The Sunnyvale and Nairobi startup is testing whether its software can coordinate computing across spacecraft built by different companies.

• Ridgeline Raises $250M at a $1.425B Valuation

  Ridgeline has raised $250 million in a Series E round led by founder Dave Duffield, valuing the AI-native investment management platform at $1.425 billion. The company says more than $750 billion in assets are committed to its platform and plans to expand its AI, managed services, and international operations.

• Sandia reports about 90%-plus accuracy in continual-learning test

  Sandia National Laboratories reported about 90% plus overall accuracy for its AIGM Lite method in an early controlled test. The project aims to help sensor-data models learn new information without losing earlier capabilities.

• Runpod’s ARR Passes $325M as Its Developer Base Tops 1.2M

  Runpod’s annual recurring revenue has topped $325 million, the company says, up from $120 million in January 2026, as its developer base grew to more than 1.2 million. The company also named Brian Erickson, formerly CFO at Suzy and Transfix, as its chief financial officer.

• Broadcom Syndicate Starts Assembling $60B AI Financing

  Broadcom’s financing partners are preparing a $60 billion debt package for AI chips and infrastructure that would benefit Anthropic and others. Bloomberg says the proposed financing includes a $42 billion senior tranche and an $18 billion junior tranche led by Blackstone.

• Vision Models Read Indoor Space Better Than They Act on It

  A paper submitted to arXiv on October 29 compares ten visual foundation models on spatial perception, prediction and goal-directed action. The models can reveal information about indoor scenes, but offline tests found that goal-sensitive actions did not reliably translate into better progress.

• CTGT Opens Concordance to Trace Inconsistent AI Decisions

  CTGT has released Concordance, an Apache 2.0 tool for tracing why an AI application produces different decisions across repeated runs. In the company’s test, one banking manual yielded four different sets of obligations even though the model gave the same answer to identical requests.

• Apple’s Minions Turns Idle Macs Into Shared GPU Capacity

  Apple researchers describe Minions, a system that pools idle Apple Silicon Macs to run machine-learning inference tasks. The paper reports that some jobs previously queued for more than 10 hours can begin within minutes.

• Together AI Describes Polymorphic Inference for Serving Open Models as Traffic and Hardware Change

  Together AI describes Polymorphic Inference as an approach that adapts to changing models, workloads and hardware. The company says Together Compiler is one of its first projects in this research agenda and is already running on some production inference workloads.

• DigitalOcean Prices Agent Droplets at $50 and $200 a Month

  DigitalOcean introduced Agent Droplets, monthly plans that discount eligible agent compute, storage, inference and tool use. Pro costs $50 a month and Team costs $200, with third-party model use and spending beyond plan allowances billed separately.

• Trust3 AI Syncs Databricks and Snowflake Policies to OneLake

  Trust3 AI announced a service to synchronize row- and column-level access policies from Databricks and Snowflake with Microsoft OneLake Security. The integration is designed to carry existing permissions into Microsoft Fabric and Power BI access paths used by people and AI agents.

• IBM and Microsoft announce Imagine Enterprise agent platform

  IBM and Microsoft announced Imagine Enterprise, a platform for building, running and governing AI agents in industry-specific processes. Its initial frameworks cover financial services, insurance and retail, and the platform is available through IBM Consulting or with self-management options.

• Philippines Seeks Comments on Draft AI Copyright Guidelines by Oct. 30

  The Intellectual Property Office of the Philippines is accepting public comments on draft guidelines for registering AI-related works through Oct. 30, 2026. The proposal would tie protection to original human expression and leave questions about AI training data outside the registration process.

• Musk Says Grok Bot’s Serving Costs Were Unsustainably High

  Elon Musk said Grok Bot’s serving costs were unsustainably high relative to revenue after a user asked whether usage could be reset. His reply included no cost or revenue figures, and xAI’s announcements list the plans that include the agent without explaining the economics of serving it.

• Andreessen Horowitz Opens a Tuition-Free AI Academy in San Francisco

  Andreessen Horowitz is backing the Horowitz Andreessen Academy, a San Francisco program for young builders entering technology after high school. The tuition-free founding fellowship begins in fall 2027 and will pair AI-focused coursework with startup projects, company placements and access to computing resources.

• Ollama Adds Gradium’s Stable Phonon Checkpoints in v0.40.9

  Ollama v0.40.9 adds support for Gradium’s GA-format Phonon speech checkpoints on Apple Silicon, replacing the earlier prerelease checkpoint path. The release also enables speech generation from the CLI and through an OpenAI-compatible API endpoint.

• Hyundai and Kia to Show Factory AI Agents at E-FOREST Tech Day

  Hyundai Motor and Kia will present factory-focused AI agents and about 200 manufacturing technologies at E-FOREST TECH DAY 2026 in Uiwang, South Korea. The three-day event will showcase digital twins, automation systems and the companies’ plans for software-defined factories.

• Meta Must Name Staff Behind Shadow-Library Downloads

  A federal judge ordered Meta to identify people who approved downloads from shadow libraries such as LibGen and Anna’s Archive. The discovery order also requires details about BitTorrent use, including decisions about seeding and efforts to limit it.

• Hedra says Model Factory will train and deploy models for real-world uses

  Hedra says its Model Factory platform combines post-training, GPU capacity and inference services for teams building models for physical-world applications. The company also reported faster image editing and results from tests involving video and a simulated robot.

• StudentSim Builds Simulated Students for AI Tutor Training

  A new arXiv paper presents StudentSim, a framework that turns sparse learner data into individualized student simulators for evaluating and training AI tutors.

• UK Links Faster Grid Access to Strategic Data Centres

  The UK’s Data Centre Covenant sets out faster grid access for strategic data centres that meet standards on security, local investment, jobs and resource use. The government says it needs 6 GW of AI data-centre capacity by 2030, but has yet to define which projects qualify for support.

• Study Finds a Mathematical Limit to AI Reproducibility

  A study proposes that intertwined training basins can make deep-learning outcomes difficult to predict even when external randomness is controlled. Its tests find near-zero uncertainty exponents in a minimal model and in deep networks, meaning finer initialization precision brings only marginal gains in predictability.

• StableVQ Proposes Training Changes for More Stable Visual Tokenizers

  StableVQ is a paper about training stability in vector-quantized visual tokenizers, the systems that convert images into discrete visual representations for generative models.

• Ai2’s StateBench Finds Long-Context Tracking Still Trips Up Models

  Ai2’s StateBench benchmark finds that frontier models struggle with harder state-tracking tasks as inputs grow, while recurrent and hybrid models generalize farther when trained directly on the tasks. The results vary by task difficulty and format, and the benchmark leaves open questions about tracking newly introduced entities.

• AWS and Cognition Sign Multi-Year Devin Deployment Agreement

  Amazon Web Services and Cognition have signed a multi-year agreement to help enterprises deploy Devin autonomous software engineers on AWS. The collaboration targets legacy-system modernization, security backlogs and parallel engineering work inside dedicated customer environments.

• OpenAI Pledges $1 Billion for Frontline Cyber Defenders

  OpenAI is committing $1 billion in subsidized Daybreak access, training and technical support for organizations that protect water, electricity, public services and other essential systems. The initiative expands to government agencies through a separate agreement that gives verified public-sector defenders discounted access to OpenAI’s cyber tools.

• Ofcom Narrows X Probe to Illegal-Content Duties

  Ofcom has removed three child-safety duties from its investigation into X after the company changed how it restricts pornography for accounts that have not completed age assurance. The regulator will continue examining X’s illegal-content duties and its handling of related complaints and reports, while separately monitoring restrictions on Grok.

• Mantic Raises $25 Million After AI Beats Human Forecasters

  London AI startup Mantic raises a $25 million seed round led by Radical Ventures after finishing ahead of every human participant in the Summer 2026 Metaculus Cup. The company specializes frontier AI models for probabilistic forecasts about politics, economics, culture and other real-world events.

• OpenAI's Chris Lehane Says Company Has Discussed AI Safety With Anthropic, Google DeepMind for Weeks

  OpenAI global policy chief Chris Lehane said the company has been discussing AI safety with Anthropic and Google DeepMind for several weeks, as Bloomberg first reported.