A Guth Labs publication ↗
Guth News

Research

Agents

Perplexity lets users set up Teammates from the Computer prompt box

Perplexity users can now start setting up a Teammate from the prompt box with Computer mode turned on, CEO Aravind Srinivas said on November 20. In the announcement captured by Minutes of AI, Srinivas said setup can use context from the current se…

Sources: x.com, perplexity.ai +1
  1. Research

    nference and Cedars-Sinai plan AI platform for clinical research

    nference announced a long-term partnership with Cedars-Sinai on November 20 to build an AI-enabled platform for clinical researchers to analyze de-identified patient records. In its announcement, nference said the platform would draw on records fr…

    Sources: prnewswire.com
  2. Research

    Wenge adds Google and Apple sign-in to ScienceOra

    Wenge AI added Google and Apple sign-in to the international site for ScienceOra, its AI research platform, the company announced on November 20. The update gives researchers another way to register or access an account without filling out the pla…

    Sources: prnewswire.com
  3. Models

    Tencent releases the weights for its flagship Hy4 model

    Tencent released downloadable weights for its flagship Hy4 model on November 20, following its hosted launch earlier in the month. The model card lists an instruction-tuned model, a reduced-precision FP8 version of that model, and a pretrained bas…

    Sources: huggingface.co, tencent.com
  4. Research

    Sakana AI launches a research group for AI security and defense

    Sakana AI announced its Security Research Group on November 20 to study how to secure AI systems and how organizations might use AI to coordinate cyber defense. The Tokyo-based team is led by Akihiro Umegai and David Ha. Sakana says the group will…

    Sources: sakana.ai
  5. Agents

    alphaXiv opens a preprint server for AI agents

    alphaXiv opened a public preprint server for AI agents after a 12-week experiment in which 936 agents produced 936 studies, according to the project’s announcement and paper. The study involved 35 human-operated labs. People supplied computing res…

    Sources: alphaxiv.org
  6. Tools

    India’s BHASHINI and NIELIT agree to explore multilingual AI for technical education

    India’s Digital India BHASHINI Division and the National Institute of Electronics and Information Technology (NIELIT) exchanged a memorandum of understanding on November 19 to explore using multilingual AI in technical education, training and publ…

    Sources: pib.gov.in
  7. Models

    Google selects Blue Origin for Project Suncatcher launch in 2029

    Google and Blue Origin announced a launch-services agreement on November 19 for Project Suncatcher, Google’s research effort to put AI computing infrastructure in space. The mission is targeted for the first quarter of 2029 and will use Blue Origi…

    Sources: blog.google, blueorigin.com
  8. Agents

    Cision connects Brandwatch research agents to external tools through MCP

    Cision announced the first release of Cision MCP on November 19, connecting supported tools and agents to Brandwatch Consumer Research capabilities. The company’s announcement says the integration uses Model Context Protocol, a standard for connec…

    Sources: prnewswire.com
  9. Agents

    Lianlian details XWorker agents and permission-controlled payments

    Lianlian DigiTech announced the formal launch of XWorker at Singapore FinTech Festival on November 19, describing a platform that coordinates AI agents with business tools and payment services. The company announcement links to an access-request p…

    Sources: prnewswire.com
  10. Tools

    Alibaba.com opens free Accio sourcing access to European SMEs

    Alibaba.com opened free access to Accio Sourcing Expert in Auto Mode for European small and medium-sized businesses on November 19, alongside a European SME Startup Pack. The announcement came at the opening of CoCreate Europe in London. The sourc…

    Sources: prnewswire.co.uk
  11. Models

    BMO reports quantum feature-selection results using IBM computers

    BMO reported on November 19 that it had demonstrated quantum utility in a machine-learning feature-selection application using IBM quantum computers, according to the bank's announcement. Developed with Democritus University of Thrace, the work us…

    Sources: prnewswire.com
  12. Research

    Isomorphic Labs and MSD partner on AI-designed oncology drugs

    Isomorphic Labs and MSD announced a research collaboration on November 19 to discover small-molecule drug candidates for an oncology target, according to their joint announcement. Isomorphic will use its proprietary drug-design engine, IsoDDE, to …

    Sources: prnewswire.com
  13. Guth News

    Youdao reports lower revenue and higher operating profit amid AI push

    Youdao reported November 19 that third-quarter revenue fell 12.3% year over year while operating profit rose 19.2%, as the learning and advertising company continued promoting its speech and translation AI strategy. Its unaudited earnings release …

    Sources: prnewswire.com
  14. Agents

    Cloudflare adds file attachments to Agent Lee

    Cloudflare added image, PDF and text-file attachments to Agent Lee on November 18, letting users supply screenshots, logs and configuration files alongside troubleshooting questions. The official changelog says the assistant remains in beta. Users…

    Sources: developers.cloudflare.com
  15. Research

    LATO raises $4.7 million for AI-led investment research

    LATO has raised $4.7 million in seed funding co-led by Air Street Capital and Axiom Partners, according to November 19 reporting by EU-Startups, to expand its AI-led commercial research platform. The startup was founded by Tymek Staniszewski and T…

    Sources: eu-startups.com, ycombinator.com +1
  16. Infrastructure

    MegaRobo raises nearly $200 million for AI-driven laboratory infrastructure

    MegaRobo announced nearly $200 million in financing in a November 18 company release, with participation from Oman Investment Authority, to expand infrastructure connecting AI with laboratory experiments and other physical workflows. Existing inve…

    Sources: prnewswire.com
  17. Agents

    OpenAI reports uneven gains from coding agents in its research work

    OpenAI published an internal research-acceleration report on November 18, describing rising coding-agent use alongside substantial differences in autonomous task performance. The company estimates that success without human intervention on researc…

    Sources: openai.com
  18. Research

    Ten63 launches BEYOND-Alpha drug-design program with Chugai

    Ten63 Therapeutics launched BEYOND-Alpha on November 18, naming Chugai Pharmaceutical as the first participant in its early-adopter program for computational drug design, according to its announcement. Selected pharmaceutical companies will work w…

    Sources: prnewswire.com
  19. Research

    Zapata and University of Maryland introduce ForShor AI research challenge

    Zapata Quantum announced ForShor.ai on November 18, an open challenge developed with the University of Maryland for AI-assisted quantum algorithm research. The project is built around a formally verified implementation of Shor’s algorithm. Partici…

    Sources: investors.zapataquantum.com
  20. Models

    Palo Alto Networks agrees to buy Applied Compute for $4.5 billion in stock

    Palo Alto Networks agreed on November 18 to acquire Applied Compute for $4.5 billion in stock, subject to adjustments, according to its announcement and securities filing. The deal is not complete. It requires customary closing conditions, includi…

    Sources: investors.paloaltonetworks.com, sec.gov
  21. Agents

    StepSecurity raises $20 million and opens a free package-security tier

    StepSecurity announced a $20 million Series A on November 18 and launched a free community tier for Secure Registry, its package-security service. Elephant led the financing, with Runtime Ventures, Inner Loop Capital and SaaS Ventures participatin…

    Sources: prnewswire.com
  22. Agents

    Google opens a waitlist for Stitch Loop, its product-maintenance agent

    Google introduced Stitch Loop on November 18, opening a global waitlist for an experimental agent that works on software after launch. Eligible Google AI Ultra subscribers receive priority access; the announcement is not a general-availability rel…

    Sources: blog.google
  23. Models

    Google brings Genie 4 to AI Ultra subscribers in Project Genie

    Google began rolling out Genie 4 in Project Genie to eligible Google AI Ultra subscribers on November 18, alongside a creation agent and community gallery, according to its announcement. The browser-based experience lets users explore generated wo…

    Sources: blog.google
  24. Policy

    Qodo previews policy-based code approvals and coding-session reviews

    Qodo introduced Auto Approval Workflows in beta and Agent Session Review as a research preview on November 18, adding policy-based approvals and a broader view of coding agents’ work, according to its announcement. Teams can set approval criteria …

    Sources: qodo.ai
  25. Research

    Tesorai raises $5.3 million to make raw research data usable for AI

    Tesorai announced $5.3 million in seed funding on November 18 to expand AI tools that analyze raw life-sciences experiment data, bringing its total funding to $8 million. Investors include Perplexity Fund, IQT, South Central Ventures and five othe…

    Sources: prnewswire.com, siliconangle.com
  26. Agents

    Strike Graph announces AI-built compliance integrations with human review

    Strike Graph announced Integration AI on November 18, offering customers an agent-assisted way to request new connections for collecting compliance evidence. The system researches a software vendor's public API documentation and generates integrat…

    Sources: finance.yahoo.com, help.strikegraph.com
  27. Robotics

    Humyn Labs raises $47 million in equity and debt for robotics data

    Humyn Labs announced $47 million in financing on November 18 to expand its physical-AI research, training-data and deployment infrastructure. The package consists of $26 million in equity led by Accel and $21 million in venture debt led by Lightho…

    Sources: humynlabs.ai, moneycontrol.com
  28. Research

    Investing.com adds multi-step research to WarrenAI

    Investing.com introduced a redesigned WarrenAI on November 18, adding multi-step financial research that can plan an investigation, gather live data and revisit gaps before returning an analysis. The company says the system can perform up to ten r…

    Sources: prnewswire.com
  29. Agents

    Oracle opens clinical-trial AI agents to early adopters

    Oracle introduced Life Sciences AI Agents on November 18, making an AI-assisted clinical-trial operations experience available to early adopters. The initial scope covers portfolio monitoring, trial planning and trial conduct. Oracle says the agen…

    Sources: prnewswire.com
  30. Models

    Bajaj Finance and Pennant invest about $4 million in Soket AI

    Bajaj Finance and Pennant Technologies invested approximately $4 million combined in Soket AI, according to a November 18 announcement from Bajaj Finserv. Bajaj Finance contributed approximately $2 million through the group's Finserv Intelligence …

    Sources: aboutbajajfinserv.com
  31. Agents

    Wiz reports malicious Microsoft Agent Framework packages on PyPI

    Two malicious versions of Microsoft's agent-framework Python package appeared on PyPI on November 17 and were subsequently quarantined, according to a Wiz Research investigation. Version 1.26.1 was uploaded at 20:35 UTC, followed by 1.26.2 around …

    Sources: wiz.io
  32. Research

    Google releases HelpBench to test AI security advice

    Google released HelpBench, a 450-question benchmark for testing AI advice about everyday privacy, safety and security, in a November 17 announcement describing research with University College London. The benchmark pairs questions with expert-writ…

    Sources: blog.google, github.com
  33. Models

    Remanence raises $30 million to train models on AI agent traces

    Remanence raised $30 million in seed funding to help companies train AI models on the execution records generated by their agents, according to EQT Ventures' November 17 announcement. EQT Ventures and FirstMark co-led the round. The Paris-based st…

    Sources: stories.eqtventures.com, zeroprime.vc
  34. Business

    LanceDB raises $57.5 million to expand its AI training data platform

    LanceDB announced a $57.5 million Series B on November 17 to expand its platform for organizing and preparing multimodal AI training data. Bessemer Venture Partners and Innovation Endeavors led the round, bringing total funding to $98.5 million, a…

    Sources: lancedb.com, siliconangle.com
  35. Models

    Model ML raises $250 million for financial-services AI agents

    Model ML announced a $250 million Series B on November 17 at a $1.5 billion valuation, financing further development of AI agents for financial and professional services. General Atlantic and QED Investors co-led the round, which brings total fund…

    Sources: generalatlantic.com, techcrunch.com
  36. Models

    AssemblyAI introduces STT-3.7 models for live conversations, agents and recordings

    AssemblyAI introduced its STT-3.7 transcription family, with separate models for live conversations, voice agents and recorded audio, in an announcement now dated November 18. All three support 32 languages with automatic detection. The Realtime m…

    Sources: assemblyai.com
  37. Research

    Calif discloses AI-assisted discovery of Java4Shell flaw

    Security firm Calif disclosed Java4Shell on November 17, saying AI-assisted research uncovered a memory-corruption flaw in OpenJDK that can enable remote code execution in affected applications. In its technical account, researcher Quang Luong des…

    Sources: calif.io, oracle.com
  38. Tools

    Reflection limits Laser security-testing tool to vetted organizations

    Reflection launched a Trusted Access Program on November 17, giving selected commercial partners access to Laser, its autonomous security-testing tool, as a research preview. The company announcement says Laser is intended for authorized testing o…

    Sources: reflection.ai, axios.com
  39. Models

    GitHub says AI security pilot produced 99 checked reports and 55 fixes

    GitHub and Microsoft reported on November 17 that an AI-assisted security pilot produced 99 privately submitted vulnerability reports across six open-source projects, with maintainers fixing 55 by November 15. Fourteen fixes had reached a release,…

    Sources: github.blog
  40. Agents

    Dreamteam raises $53 million for a CRM built around four AI coworkers

    Dreamteam announced $53 million in funding on November 16 to expand customer relationship management software built around shared customer context and specialized AI agents. A company post dated November 17 says the total spans its seed and Series…

    Sources: dreamteam.ai, globenewswire.com
  41. Models

    HeyGen launches Avatar 6 with script-aware gestures and two-person scenes

    HeyGen launched Avatar 6 on November 17, adding gestures that respond to a script and support for two avatars speaking in the same scene. Its product page describes controls for emotion, expression and movement, alongside API and MCP access. The c…

    Sources: heygen.com
  42. Agents

    Knotch launches Ace for Agents to make website content readable by AI assistants

    Knotch launched Ace for Agents on November 17, adding a service that prepares companies’ existing website content for AI search and personal assistants without replacing the pages human visitors see. The product sits within its Agentic Content Eng…

    Sources: prnewswire.com
  43. Tools

    Elementl uses Google Gemini to plan nuclear projects

    Elementl Power announced a collaboration with Google Cloud on November 17 to build AI tools for developing nuclear power projects. Its Lattice project-delivery suite uses Gemini for business and BigQuery, according to the companies’ announcement. …

    Sources: googlecloudpresscorner.com, prnewswire.com
  44. Agents

    Ordway adds lockbox check matching to its cash application agent

    Ordway expanded its cash application agent on November 17 to match bulk bank lockbox deposits with invoices. The company’s announcement says the update analyzes individual check images and bank lockbox reports to reconcile payments that may appear…

    Sources: prnewswire.com
  45. Research

    Dell’Oro forecasts $42 billion data-center power market by 2030

    Dell’Oro Group forecasts that the global market for high-density data-center power equipment will exceed $42 billion in 2030 as AI infrastructure moves toward higher-voltage power distribution. The research firm published the forecast on November …

    Sources: prnewswire.com
  46. Business

    Akash announces fixed-price GPU marketplace for 2027

    Akash announced a marketplace for buying future GPU capacity at fixed prices on November 17, with a pilot for certified participants planned for the first quarter of 2027. The product, called Akash Forwards, is designed around contracts that settl…

    Sources: businesswire.com, forwards.akash.network
  47. Models

    Noeri launches with $110 million to build enterprise AI that keeps learning

    Noeri, founded by Stanford professors Yejin Choi and Carlos Guestrin, launched publicly on November 16 with $110 million in funding. Menlo Ventures led the round, with investors including Madrona, NVIDIA, Cisco Investments and Snowflake. The finan…

    Sources: businesswire.com, noeri.ai +1
  48. Business

    Flagship commits £100 million to a London lab for creating AI-native companies

    Flagship Pioneering announced on November 16 that it is establishing a London company-creation lab backed by an initial £100 million commitment. The new U.K. operation will develop AI-native platform businesses in health and other scientific field…

    Sources: flagshippioneering.com, streetinsider.com
  49. Models

    Shanghai AI Lab opens Sage S2.1 Flash for scientific workflows

    Shanghai Artificial Intelligence Laboratory has made Sage S2.1 Flash available with downloadable weights and an FP8 variant, adding an openly licensed option for research workflows that combine scientific data, coding and tool use. The lab’s offic…

    Sources: github.com
  50. Tools

    LVMH’s media division pairs newsroom AI tools with ChatGPT distribution

    LVMH’s media division and OpenAI announced a strategic agreement on November 16 covering both internal AI use and the presentation of the division’s journalism in ChatGPT. The announcement published by OpenAI names ChatGPT Enterprise and other Ope…

    Sources: openai.com
  51. Guth News

    AMD confirms World Labs acquisition close as Fei-Fei Li joins its leadership

    AMD announced on November 16 that it had completed its acquisition of World Labs. The company’s announcement says Fei-Fei Li is joining AMD as executive vice president and chief scientist, reporting to Lisa Su. The announcement date and closing da…

    Sources: newsroom.amd.com, ir.amd.com
  52. Research

    Fermilab, IBM and NVIDIA plan a modular quantum research platform

    Fermilab, IBM and NVIDIA announced plans for a quantum-computing co-design platform on November 16. The proposed effort would combine quantum processors and accelerated classical computing to investigate error correction, interconnected architectu…

    Sources: nvidianews.nvidia.com, news.fnal.gov
  53. Research

    OpenAI expands researcher grants, with applications opening November 23

    OpenAI announced an expansion of ChatGPT for Academic Researchers on November 16, raising the program’s overall target to 200,000 researchers worldwide. The next round opens November 23 and offers 50,000 grants to eligible applicants on a first-co…

    Sources: openai.com
  54. Models

    OpenAI names four organizations for embedded safety assessments

    OpenAI said on November 16 that it has signed agreements with Apollo Research, METR, Transluce and Goodfire for embedded assessments of model alignment and safeguards. The program gives outside assessors scoped access to internal systems and evide…

    Sources: openai.com
  55. Models

    NVIDIA packages Nemotron models and chemistry data for scientific agents

    NVIDIA introduced Nemotron for Science at SC26 on November 16, combining its Nemotron 3.5 Super VL model with scientific evaluations, chemistry data and post-training recipes. The company says the model and LitmusBench v0.2 dataset are available o…

    Sources: nvidianews.nvidia.com
  56. Tools

    Oak Ridge opens Lux to scientific users as Genesis Mission gains computing capacity

    Oak Ridge National Laboratory’s Lux supercomputer opened to scientific users on November 16, adding dedicated capacity for AI-driven research. The laboratory reports that Lux debuted at No. 14 on the latest TOP500 list with 220.79 petaflops of per…

    Sources: olcf.ornl.gov, newsroom.amd.com +1
  57. Agents

    Superfluid research targets the queue between local agents and their human user

    guthlabs.newsroom | November 16, 2026 A technical report submitted November 13 describes Superfluid, a local inference server designed to keep an interactive chat responsive while agents share its computing resources. Rather than replacing model r…

    Sources: arxiv.org, github.com
  58. Models

    Craig details GPT-6 Astra’s role in a candidate lattice Standard Model

    Nathaniel Craig has described how he used GPT-6 Astra to develop a candidate lattice definition of the Standard Model. His account, published on OpenAI’s science blog on November 15, brings a research workflow into view, rather than announcing a n…

    Sources: openai.com, arxiv.org
  59. Models

    Apple opens a health-AI study focused on trust, comprehension and everyday use

    Apple announced the Apple Health AI Usage Study in the United States on November 15, pairing a virtual study and app with AI technology from OpenEvidence. The company says the research will examine user trust, response quality, comprehension and e…

    Sources: apple.com
  60. Agents

    Soteria pairs verified plans with runtime checks to keep AI agents on task

    A November 13 research paper proposes a different approach to agent guardrails: give the system a checked plan it can return to when an action is blocked, rather than merely telling it to stop. Soteria, by Mike He, Ankush Desai, Sharad Malik and A…

    Sources: arxiv.org
  61. Agents

    E4S 26.11 brings agent tools and browser desktops to scientific-computing containers

    E4S is bringing more AI development tools into its scientific-computing software stack. A publicly accessible release announcement dated November 16 says version 26.11 is available with more than 125 HPC and AI packages. The project lists Hermes A…

    Sources: e4s.io
  62. Agents

    SciMatrix 1.5 puts researchers inside multi-agent scientific discussions

    The Chinese Academy of Sciences’ Computer Network Information Center announced SciMatrix 1.5 on November 11, adding human intervention, selectable agent discussion rooms and a pilot collaborator-recommendation feature. This is newly covered here, …

    Sources: www1.cstcloud.cn
  63. Research

    Human Protein Atlas releases version 26, highlighting AtlasAI and new biological resources

    The Human Protein Atlas has released version 26, pairing an expanded biological reference database with AtlasAI, an assistant designed to search, analyze and explain its data. The November 11 update adds four resources covering metabolism, protein…

    Sources: proteinatlas.org, assets-eu.researchsquare.com
  64. Models

    Hugging Face publishes a recipe for shrinking multi-vector search

    Hugging Face published a practical guide on November 11 showing how retrieval models can learn to preserve useful information when their document representations are compressed. Tom Aarsen's Sentence Transformers example builds on earlier pooling …

    Sources: huggingface.co, sbert.net +1
  65. Models

    Hedra’s Model Factory pairs research agents with human-controlled deployment

    Hedra launched Model Factory on November 10, offering teams a managed way to adapt and run models for robotics, manufacturing and other physical-world applications. Its announcement describes GPU capacity, post-training and deployment as parts of …

    Sources: globenewswire.com, hedra.com +1
  66. Policy

    French panel calls for pre-launch checks on consumer AI

    A report submitted to President Emmanuel Macron proposes stronger oversight of chatbots, including scrutiny of addictive design and hidden commercial influence. Guth News | Policy | Story date: November 10, 2026 An expert commission convened by Fr…

    Sources: elysee.fr, conseil-ia-numerique.fr +1
  67. Agents

    Graphon raises $35 million to expand memory infrastructure for AI agents

    Graphon Labs announced a $35 million Series A on November 10, led by RRE Ventures with participation from HP, Energy Impact Partners, Hitachi Ventures and Gaia VC. The round brings announced funding to $43.3 million. Company announcement The compa…

    Sources: businesswire.com, thenextweb.com
  68. Research

    Google researchers use AI and Lean proofs to reduce quantum-program resource costs

    Researchers at Google DeepMind and Google Quantum AI describe an AI-assisted method for designing quantum programs whose correctness and resource costs are checked with the Lean theorem prover. The preprint first appeared November 6 and was revise…

    Sources: arxiv.org, borowiecki.dev
  69. Agents

    Agent memory can carry harmless preferences into dangerous contexts

    Persistent memory is supposed to make an assistant more useful. A November 5 preprint asks what happens when it remembers a preference but loses the limits that made it reasonable. In The Memory Trap, researchers compare six frontier models with a…

    Sources: arxiv.org, github.com
  70. Models

    WorldParticle reuses one AI architecture across materials, while training separate models

    MIT highlighted WorldParticle on November 9, describing research from MIT, the University of British Columbia and collaborators into a shared architecture for simulating materials. The project covers phenomena ranging from moving cloth and sand to…

    Sources: csail.mit.edu, arxiv.org
  71. Models

    Tether introduces BotPsy, a small open-weight model for local tool use

    Tether AI Research introduced BotPsy on November 11, a 0.8-billion-parameter model based on Qwen3.5-0.8B and specialized for tool calling. Its team reports stronger results than its base model across agent benchmarks, including a BFCL v4 score of …

    Sources: huggingface.co
  72. Tools

    Sumary raises $25 million to expand AI accounting software across Europe

    Sumary announced a $25 million seed round led by Singular on November 10, with existing backers byFounders, Partech and Tenity participating. The accounting-software company plans to expand its commercial, implementation and research teams, with B…

    Sources: sumary.com, techfundingnews.com
  73. Research

    Sage-Taiyi research puts revisable scientific assumptions at the center of AI discovery

    A new preprint introduces Sage-Taiyi, a research-agent architecture that revises its working assumptions as evidence accumulates and uses those revisions to choose what to investigate next. The paper was submitted on November 9 at 17:38 UTC. arXiv…

    Sources: arxiv.org, github.com
  74. Robotics

    Robbyant releases LingBot-Touch to help robot hands share tactile understanding

    Robbyant, Ant Group’s embodied-AI company, announced LingBot-Touch on November 10, offering a tactile perception layer intended to work across different robot hands and control models. The release targets an integration problem: touch sensors base…

    Sources: via.tt.se, arxiv.org +2
  75. Agents

    CapLite puts small-agent write permissions in code, with an explicit utility cost

    A November 12 preprint proposes a defense for small open-weight agents that does not ask their language model to author the security rules. CapLite derives policy from tool schemas, fixes write operations before reading tool results and checks con…

    Sources: arxiv.org
  76. Agents

    PRISM tests which tools an agent actually needs before granting access

    A November 8 preprint introduces PRISM, a method for reducing an agent's tool privileges before execution. It tests tool removals in resettable sandboxes, learns which capabilities a task needs and groups tasks into reusable roles. This is newly r…

    Sources: arxiv.org
  77. Agents

    Mirendil raises $1 billion for agent-led AI research, with product proof still the test

    Mirendil announced a $1 billion Series A on November 12, led by Kleiner Perkins and Clearlake Capital. Its announcement lists Google among participating investors and describes Lor, its agent-based research platform, as a private beta. Company ann…

    Sources: mirendil.com, prnewswire.com +1
  78. Research

    OpenAI–Upwork study finds larger AI project budgets, not a comparable hourly-rate premium

    OpenAI and Upwork published research on November 10 suggesting that AI is changing how freelance projects are assembled. The evidence points to different combinations of skills and larger assignments, rather than a straightforward story of either …

    Sources: openai.com, cdn.openai.com
  79. Models

    Corelight finds hidden dependency changes beneath pinned MCP servers

    Corelight has documented a deployment risk in AI assistants connected to administrative tools: a server can keep the same release identifier while its supporting code changes. The company’s November 10, 2026 study concerns installation choices, no…

    Sources: corelight.com, docs.astral.sh
  80. Models

    OpenAI study finds training gains can make reasoning traces harder to monitor

    OpenAI’s November 10 research report examines how model training affects the usefulness of written reasoning traces for safety oversight. The company says observations from GPT-6 Astra prompted the study. OpenAI research The researchers report tha…

    Sources: alignment.openai.com
  81. Agents

    OpenAI opens Business Hub and customer-facing agents inside ChatGPT

    OpenAI introduced Businesses in ChatGPT in beta on November 10, giving companies a way to claim profiles, update their information and inspect aggregate engagement data. Eligible businesses can use Business Hub wherever ChatGPT is available; Busin…

    Sources: openai.com
  82. Agents

    NIST opens an agent-focused attack taxonomy for public comment

    NIST has opened public consultation on a revised adversarial-machine-learning taxonomy that puts attacks against agentic systems at the center of its latest edition. Released November 9, the initial draft of SP 2400-102e2026 is accepting comments …

    Sources: nist.gov, csrc.nist.gov
  83. Models

    New York expands independent AI testing and opens frontier-developer registration

    New York announced a statewide AI-testing network on November 10 and opened registration for large frontier developers ahead of January compliance requirements. The initiative joins technical scrutiny with a concrete reporting process. The New Yor…

    Sources: suny.edu
  84. Agents

    Commvault plans governed AI data access with NVIDIA and Databricks

    Commvault announced two AI-data integrations on November 10, aiming to turn information it already protects into material that enterprise applications and agents can retrieve without losing permissions or context. Both are planned releases, not br…

    Sources: ir.commvault.com, commvault.com
  85. Agents

    n8n releases Nod, a local personal agent with open-source code

    n8n has introduced Nod, a standalone personal AI agent for Mac and Windows, as a research preview. Announced November 10, the app comes from the new n8n Labs team and is separate from the company’s workflow-automation product. Users connect their …

    Sources: blog.n8n.io, github.com
  86. Tools

    GRAIL opens a €1 million call for robotics data and evaluation tools

    Europe’s GRAIL project opened its Genesis call on November 9 to support the data and evaluation infrastructure behind physical AI. The call offers a €1 million budget and up to €100,000 for each selected project, with applications closing February…

    Sources: grail-robotics.ai, exteriors.gencat.cat
  87. Research

    InfraRSI research targets faster AI training through coordinated infrastructure changes

    A research team has introduced InfraRSI, an agent-based system that searches for ways to accelerate distributed language-model training by changing launch settings and code together. The preprint was submitted to arXiv on November 8. Paper. Rather…

    Sources: arxiv.org, github.com
  88. Models

    IFM releases PAN 2 world model with training code and open weights

    The Institute of Foundation Models at Mohamed bin Zayed University of Artificial Intelligence announced PAN 2 on November 10, releasing a world-model system intended to simulate the consequences of actions before an agent takes them. IFM says its …

    Sources: ifm.ai, github.com
  89. Models

    MMDM exposes the gap between choosing one answer and choosing every valid action

    A new benchmark asks a practical question about single-pass decision models: can they identify all valid choices when the options depend on one another? MMDM, submitted to arXiv on November 10, evaluates 28 models across 5,448 questions in 16 cate…

    Sources: arxiv.org
  90. Research

    Adaspace listing gives orbital AI computing a public-market test

    AI infrastructure is seeking capital beyond terrestrial data centers. Adaspace began trading in Hong Kong on November 13, according to its listing announcement and independent coverage. The issuer's earlier HKEX offering notice identifies stock co…

    Sources: www1.hkexnews.hk, philpr.com +1
  91. Agents

    World details Cloudflare’s agent-trust push, but local agents remain outside its initial scope

    World’s November 12 explanation of Trusted Agent Delegation gives agent builders a practical boundary to watch: the initiative initially focuses on operator-hosted agents, not on-device or locally hosted ones. Its goal is to help websites assess t…

    Sources: world.org, blog.cloudflare.com +1
  92. Robotics

    T-Rex separates robot planning from fast touch corrections

    guthlabs.newsroom · November 12, 2026 Panasonic's November 5 research announcement highlights a robot-control approach that gives touch its own faster feedback loop. T-Rex combines vision-and-language motion planning with tactile corrections, aimi…

    Sources: news.panasonic.com, arxiv.org +1
  93. Research

    OpenAI pledges $2 million for child-safety research, with evidence still to come

    guthlabs.newsroom · November 12, 2026 OpenAI announced on November 12 that it has committed $2 million to research on AI and children, following its signing of a child-safety charter earlier this week at the Paris Peace Forum. The pledge supports …

    Sources: openai.com
  94. Models

    Cloudflare brings multimodal EmbeddingGemma 2 to Workers AI

    November 12, 2026 | guthlabs.newsroom Cloudflare has added Google's EmbeddingGemma 2 to Workers AI, giving developers a hosted embedding option for mixed collections of text, code, images, video and audio. The announcement is a new Cloudflare avai…

    Sources: developers.cloudflare.com
  95. Models

    MotionGraphicsBench tests whether AI animations actually follow the brief

    November 7, 2026 | guthlabs.newsroom Researchers affiliated with Figma and Stanford have introduced MotionGraphicsBench, a benchmark of 100 tasks that ask language models to animate supplied static SVG artwork. Providing the design upfront separat…

    Sources: arxiv.org, huggingface.co