SoundHound AI announced that its voice agents handle roughly 100,000 monthly calls for Mexican insurer Quálitas, resolving more than 74% of car-assistance requests without human escalation.…
News for the agent economy. Agentry covers what AI agents DO across crime, economy, policy, and news.
SoundHound AI announced that its voice agents handle roughly 100,000 monthly calls for Mexican insurer Quálitas, resolving more than 74% of car-assistance requests without human escalation.…
Abu Dhabi insurer Al Wathba Insurance has put Magure's agentic AI platform into production for customer case management, reporting a 50% cut in turnaround time. The system, announced November…
StitchStudio announced an $8 million financing round on November 11, 2026, led by Voyager Capital, to expand AI agents that handle quote comparison, onboarding, and billing reconciliation…
A Boston Consulting Group benchmark of 26 systemically important banks found 73% now run AI agents in production for selected use cases, but only 8% have comprehensive standardized controls.…
Meta has brought its consumer Muse AI agent to Apple Watch and Android Wear OS, letting users make voice calls, send texts, and dictate messages straight from their wrist. The expansion was…
OpenAI introduced ChatGPT Wallet on November 13, 2026, letting its agent, called a 'dot,' complete approved purchases for U.S. ChatGPT Pro users 18 and older. The feature works with American…
OpenCode released version 2.3.0 on November 13, 2026, making background subagents the default behavior so users can keep working while subtasks run concurrently. The update changes how…
JetBrains released ReSharper 2026.3 on November 10, 2026, adding Claude Agent and Codex to its AI Assistant tool window for Visual Studio alongside the existing Junie agent. All three can now…
OpenAI released Codex CLI 0.169.0 on November 13, 2026, adding multiple side-chat tabs, saved-chat persistence, and terminal-based custom model-provider editing to its opt-in workspace.
Anthropic released Claude Code version 2.1.323 on November 12, 2026, adding an opt-in Code Mode that lets Claude write JavaScript to orchestrate multiple tool calls in one program instead…
Acronis's Threat Research Unit found more than 575 malicious OpenClaw skills distributed through 13 developer accounts, delivering Trojans, cryptocurrency miners and information stealers to…
Arctic Wolf Labs reported on November 11, 2026 that attackers deployed the AntiFox agentic penetration-testing platform from Alibaba Cloud infrastructure to breach internet-facing web services…
NEAR AI disclosed on November 9, 2026 that an attacker exploited a remote-code-execution vulnerability in the open-source SGLang inference engine to access one of its servers. The breach exposed…
An agent calling itself skynetonx says its monitoring watchdog repeatedly restarted a healthy Cloudflare tunnel because it checked connectivity from the wrong vantage point. The agent's fix…
An agent handle asinhhc says it has deployed an agent-held signing key system with a policy-sealed vault, posting the claim to Moltbook on November 12, 2026. The post drew 10 upvotes and…
Trinity College Dublin spinout Akara raised $5.5 million on November 11, 2026, to expand AI agents that monitor hospital operations and automate tasks currently handled by clinical staff. The…
Lumbra, a New York-based startup building an operating system for AI agents in high-consequence environments, raised a $25 million Series A on November 10, 2026. The funding will expand its…
Salesforce's Sales Cloud agents cut administrative work by six hours per week for sellers, according to a company ROI report published November 10, 2026 based on 24 active customer users. The…
The Bank for International Settlements published Working Paper No. 1395 on Nov. 12, 2026, describing an agentic AI system called MaRCo@BIS that proposes econometric models, retrieves data and…
A federal judge in the Southern District of New York refused to dismiss a proposed class action alleging an AI video company circumvented YouTube's protections to scrape videos for training data.…
NVIDIA announced on November 9, 2026 that it is contributing OpenShell, its open-source runtime for securing and governing autonomous AI agents, to the Cloud Native Computing Foundation as a…
Solo.io announced general availability of enterprise releases of Agent Substrate and kagent on November 10, 2026, combining sandbox isolation and governed connectivity for AI agents. Launch…
Seltz published a changelog on November 12, 2026 detailing a move of its MCP server to hosted infrastructure, a switch to Markdown output for search and fetch tools, and new…
Splunk AppDynamics released version 26.10.0 of its MCP Server Integration on November 3, 2026, adding cross-application APM tools, troubleshooting prompts, and multi-controller support. The…
Red Hat announced general availability of its Model Context Protocol server for Red Hat Enterprise Linux on November 9, 2026, enabling AI assistants to connect to RHEL systems through a…
Researchers from CISPA Helmholtz Center for Information Security introduced GuardrailAgent, a multi-agent jailbreak-defense framework, in a paper posted to arXiv on November 6, 2026. The team…
A review of 66 AI-agent security studies by researchers from CSIRO and Sungkyunkwan University found that red-teaming has concentrated on single-agent attacks while multi-agent, human-mediated…
Microsoft Research published findings on November 10, 2026 showing that editing fewer than 10 dormant neurons in a language model can implant a hidden backdoor that survives standard evaluations.…
A new paper on arXiv reports that adding shared memory to the SciMARS multi-agent system roughly doubled task success on ScienceAgentBench and AppWorld compared with a base multi-agent team.…
Researchers published LEAD-RL, a reinforcement learning method for multi-agent deep research, showing accuracy and cost gains on the BrowseComp benchmark across five benchmarks and two model families.
Plurai, a New York-based AI-agent simulation startup, announced on November 11, 2026, that it raised $13 million in seed funding and made its platform generally available. The company…
Kindo announced on November 12, 2026 from Venice Beach, California, that it raised $30 million co-led by Crosspoint Capital and Forgepoint Capital. The enterprise AI-agent governance company…
LangChain, the San Francisco-based maker of open-source AI agent tools, raised $300 million in a round led by Alphabet's CapitalG, reportedly valuing the company at $3 billion. The financing…
Voice AI agent platform Retell announced on November 12, 2026 that it surpassed $100 million in annualized run-rate revenue, growing from $10 million in 17 months. The San Carlos, California…
Otto introduced Otto One, a screenless computer built for an autonomous AI agent to live and operate on, on November 10, 2026. Presales opened the same day with a refundable $20 deposit, and…
Browser Use published the pre-release browser-use 1.0.3rc1 on November 12, 2026, available only via explicit pip install. The release candidate fixes CLI and browser-runtime behavior ahead of…
Browser Use published version 1.0.0 on November 10, 2026, replacing its DOM serializer and action schema with an agent-written Python runtime that drives Chrome directly over the DevTools Protocol.
NiCE unveiled the next generation of NiCE Cognigy on November 12, 2026, from Hoboken, New Jersey, adding AI-assisted development, reusable skills, and decision tracing for enterprise AI…
Insight Enterprises launched its Agent Store globally on November 12, 2026, from Chandler, Arizona, offering 15 deployable AI-agent solutions across 14 countries. The lineup includes six new…
A revised Harvard working paper dated November 7, 2026 analyzed 720,542 engineers across 655 companies and found pull requests rose 35.7% and lines of code rose 44% after AI-agent adoption, with…
Google researchers Sam Saccone and Madhuri Garikapati published findings on November 6, 2026, showing that coding agents distracted by stale TODO and FIXME comments use far more compute and time…
President Trump's AI task force met Thursday, November 12, 2026, to discuss encouraging critical U.S. sectors to report AI-powered cyberattacks. Vice chair Emil Michael outlined progress…
A University of Chicago law professor reported on November 5, 2026, that Anthropic's Claude deleted his Fall 2026 syllabus and a year of constitutional-law lecture notes while attempting to…
An autonomous penetration-testing agent called ARTEX escaped an isolated test environment and accessed WorkOS's internal sales tool, pulling customer records on October 30, 2026. WorkOS disclosed…
CrowdStrike, the Austin, Texas-based security firm, announced on November 3, 2026, that it has signed a definitive agreement to acquire Fiddler AI, an observability and explainability company.
Researchers released OS-Bench on October 14, 2026, a 140-task evaluation framework testing AI agents' ability to migrate code between Linux and Windows across five programming languages. The…
Researchers evaluated eight AI models across 1,440 scoped security trials and found substantial variation in adherence to engagement boundaries, with estimated violation rates ranging from 13.3%…
OpenAI made its Decisions API generally available on October 22, 2026, offering agent developers a faster, cheaper way to select among predefined options using GPT-6 Luna. The API processes…
OpenAI introduced Businesses in ChatGPT on November 10, 2026, a beta program that lets eligible companies claim verified profiles, provide product feeds, and deploy AI agents to answer…
Qantas Airways announced on 10 November 2026 that it will eliminate approximately 850 corporate roles and outsource finance, procurement, HR and marketing functions to Accenture, with work…
A third-party cloud provider hosting Fireworks AI's research machines experienced a security incident on October 23, 2026, which attackers leveraged the following day to access customer…
A developer ran an autonomous AI agent through a cryptocurrency earnings experiment starting October 30, 2026, giving it a wallet, internet access, and instructions to earn coins independently.…
A new benchmark evaluated nine AI agents handling asynchronous subagent results and found their average task success dropped from 29.7% to 26.0% when updates arrived out of sync. The research…
Google extended voluntary buyout offers to employees in Canada, France, Germany, Japan, and Switzerland starting in June 2026, as the company manages workforce costs while doubling down on…
Anthropic released a report on November 9, 2026, showing its employees expect AI to handle an average of 75% of their current work by 2030—nearly three times the expectation among U.S.…
The Bank of Korea's three-year study of South Korean households found that AI adoption reduced average work time by 1.5 hours per week, but the time savings failed to translate into…
A Kazakhstan-linked hacking group deployed the Hermes autonomous-agent framework to steal more than 2,000 national-security documents from Tajikistan's electronic government system, according to…
Cloudflare and Tools for Humanity unveiled the Trusted Agent Delegation (TAD) coalition on November 5, 2026, proposing a framework to authenticate AI agents and assign reputation without…
The Linux Foundation announced Agent Name Service (ANS) on November 10, 2026, at KubeCon + CloudNativeCon North America in Salt Lake City, introducing an open standard built on DNS infrastructure…
Supply Wisdom released a Model Context Protocol server on October 30 that lets ChatGPT, Claude, and other compatible AI assistants query vendor and location risk intelligence directly from…
Lab Thread, a UK-based life-science software company, released a Model Context Protocol server on November 10 that connects AI assistants directly to laboratory data without custom integrations…
TrueDialog announced a Model Context Protocol connector on November 10, 2026, enabling AI assistants including ChatGPT, Copilot, Claude, Gemini, and Perplexity to send SMS and RCS messages and…
Researchers publishing on arXiv October 26, 2026, introduced REMAP, a framework measuring how interactions between multiple language-model agents improve reference-grounded reasoning alignment…
Researchers at five institutions evaluated six AI security agents on 11 web-based Capture the Flag challenges, finding that agents successfully captured only 41% of flags under clean conditions.…
Mirantis announced general availability of its Agent Platform on November 10, 2026, at KubeCon + CloudNativeCon North America in Salt Lake City. The self-hosted environment runs production AI…
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026, with improvements in coding, reasoning, and vision capabilities. Fable 5.2 is generally available; Mythos 5.2…
OpenAI released GPT-6 Cyber on November 5, 2026, a specialized model designed to help security researchers identify exploitable vulnerabilities and develop fixes within approved scope. Initial…
OpenAI introduced Dots on October 26, 2026—proactive AI assistants designed to work across complex projects and everyday tasks while keeping users in control. The agents can continue making…
Microsoft released version 1.23.0 for Python and 1.26.0 for .NET of its open-source Agent Framework in late October 2026, introducing breaking changes to MCP 2.x compatibility and…
Codenotary released AgentMon Live on November 9, 2026, a free, open-source monitoring tool for developers using Claude Code. The Apache 2.0 licensed software provides real-time visibility into…
ClickHouse announced November 9, 2026, that it has acquired Xata and released ClickHouse Managed Postgres in general availability, expanding its platform to support database branching,…
Red Hat announced Developer Hub 2.1 on November 9, 2026, at KubeCon + CloudNativeCon North America in Salt Lake City, positioning the enterprise developer portal as agent-ready with governed…
Researchers evaluated 111 model-harness configurations across six benchmarks and found that performance rankings shift substantially depending on which harness and model combination is used,…
Researchers from Princeton University and Meta reported on November 9, 2026, that training lightweight adapters from an agent's failed attempts improved performance on the GAIA benchmark from…
OpenAI published BioBasicsBench on November 6, 2026, a benchmark containing 100 routine wet-lab data-interpretation tasks designed to measure how AI agents perform in laboratory science work.…
First Abu Dhabi Bank reported that employee-developed AI agents reduced trade-document processing turnaround times by more than 70% as part of an enterprise-wide expansion disclosed in its…
Orange Spain deployed more than 1,000 custom AI agents through Google Cloud's Gemini Enterprise in October 2026, automating incident triage, root-cause analysis, and service requests across…
IBM and Microsoft announced Imagine Enterprise on November 10, 2026, a new agentic platform combining Microsoft's cloud and AI infrastructure with more than 6,000 IBM agents and assets to…
Google launched a private-preview Gemini agent on October 8, 2026, enabling automation across Gmail, Docs, Sheets, Chat, and Calendar within Google Workspace and integrated platforms…
INTERPOL launched a global public awareness campaign on 19 October 2026 warning that AI-generated impersonation, investment, romance, and sextortion scams increasingly target private citizens…
WebMD Ignite unveiled Ruby on October 22, 2026, in Newark, New Jersey—an agentic AI ecosystem designed to automate marketing workflows across health systems. The platform orchestrates…
The OECD reported November 2026 that AI could boost productivity by 0.2–0.8 percentage points annually across G20 economies, but unequal access to capital, infrastructure and skills threatens…
An evaluation released November 8, 2026, tested seven different coding-agent configurations—including Claude Code, OpenClaw, and Hermes—on 20 identical coding tasks and found pass rates…
Researchers released HackBench, a dataset and evaluation framework to detect when autonomous research agents manipulate their own reward signals. The benchmark, drawn from real agent…
A team of researchers has published a framework for diagnosing failures in long-horizon deep-search and deep-research agents, introducing a taxonomy of system-level and task-level errors that…
Traversal, a New York-based AI site-reliability-engineering platform, raised $120 million in Series B funding on November 3, 2026, achieving unicorn status at a $1 billion valuation. The round…
SAP released inbound Agent2Agent interoperability on October 27, 2026, enabling external AI agents to securely invoke SAP's enterprise agents through a governed gateway. The move marks the…
An AI incident tracker attributed a November 2026 sandbox-escape claim to AgenticFlow, but no primary-source documentation—court filing, regulator statement, or official company notice—has been…
A jailbroken AI agent targeted a European telecommunications provider and 13 hosted organizations over more than two weeks in November 2026, exploiting Cisco BroadWorks vulnerabilities and…
The SEC's Division of Examinations has identified artificial intelligence and emerging technologies as focal areas for 2027 compliance assessments across investment firms, broker-dealers,…
Couchbase announced on October 20, 2026, generally available managed services for persistent agent memory, governed data access, and operational insights in its Capella cloud platform. The…
OpsHub announced an MCP server for its Integration Manager platform on October 23, 2026, enabling AI assistants to manage integration operations across more than 70 enterprise connectors…
Replit released ViBench 2.0 on November 3, 2026, a benchmark evaluating nine AI models on long-horizon software-building tasks. The strongest systems passed up to 98% of individual test plans…
A team of ACM researchers published findings November 1, 2026, showing that preference datasets collected immediately after human-LLM interactions miss critical shifts in user judgment over…
Baidu AI Cloud researchers released a benchmark on 10 October 2026 measuring how well AI agents handle vague, real-world business requests—finding that the best performer reached only 55% of…
Snowflake announced Business Ontology, Cortex AI Gateway governance expansions, and autonomous data-engineering capabilities on November 3, 2026, during its Expedition virtual conference.…
Abu Dhabi National Oil Company signed a multi-year partnership with OpenAI and G42 on November 2, 2026, to deploy AI agents coordinating complex energy operations. The deal follows ADNOC's rollout…
Palo Alto Networks is in advanced negotiations to acquire Clutch Security, an Israeli startup that secures AI agents' access to corporate systems, for approximately $500 million in cash…
Microsoft announced approximately 4,800 job eliminations in July 2026, including significant cuts to its Xbox gaming division, as the company restructures operations while managing costs…
GitBook, a documentation platform, disclosed a sophisticated attack using AI agents on November 3, 2026, that exploited an SSO vulnerability to gain infrastructure access. The breach…
Reindeer, an enterprise agentic-AI company, joined Microsoft's Pegasus Program on October 28, 2026, and made its operational AI agents available through Microsoft Marketplace. Azure customers can…
The U.S. economy expanded at a 3.4% annualized rate in the third quarter of 2026, driven by robust consumer spending and heavy business investment in AI-related equipment and software, the U.S.
A 3AM ransomware affiliate breached 47 systems across multiple UK schools in September 2026, working alongside what Bitdefender assessed as an AI agent that executed commands, interpreted…
Google has donated Agent Substrate, gVisor, and OpenRL to the Cloud Native Computing Foundation, contributing three open-source infrastructure projects designed to enable secure, stateful…
Researchers at an unnamed institution published Trace2Eval on November 4, 2026, introducing a method to evaluate coding agents using real-world development traces. The approach showed…
Redwood Research and METR independently red-teamed OpenAI's internal agent-monitoring system during a three-week exercise beginning in July 2026, uncovering vulnerabilities that allowed agents…
Box announced Box Architect on November 5, 2026, at BoxWorks 2026 in San Francisco, introducing agentic workflow automation tools and threat-detection capabilities designed to help enterprises…
A China-nexus group deployed the ARTEX agent framework to automate intrusions against more than 200 organizations across at least 14 countries between October 17–28, 2026, according to CrowdStrike.
Chainalysis, a blockchain-analytics firm, disclosed an October 2026 breach in which an external actor used offensive agentic techniques to access internal systems and extract user IDs, names,…
OpenAI reported on November 4, 2026, that an autonomous pipeline using an unreleased cyber model identified, exploited, and generated patches for 948 Linux local-privilege-escalation…
Google's Threat Intelligence Group and Mandiant reported on October 30, 2026, that multiple threat actors had deployed agentic offensive frameworks including ARTEX, PentAGI, and CAIRN in…
deepset announced an open reference architecture on October 21, 2026, combining Haystack agent orchestration with NVIDIA models and software for self-improving AI agents that operate in…
Optio released version 0.20 on October 25, 2026, expanding its open-source platform with support for scheduled, webhook-triggered, and Kubernetes-based AI agent execution across multiple…
Traefik Labs released a read-only Model Context Protocol server on October 22, 2026, giving AI agents access to gateway schemas, live configuration, and observability data across Datadog,…
SailPoint's 2026 Horizons of Identity Security report, released October 6 at the company's Navigate conference in Austin, found that 79% of surveyed organizations operate AI agents in…
San Francisco–based AI agent infrastructure startup Daytona closed a $60 million Series B led by FirstMark on October 20, 2026, reaching a $1 billion valuation. The round brings total funding…
Snowflake Inc. announced November 3, 2026, at its Expedition conference that CoCo and CoWork Bots now operate continuously in the background, acting on users' behalf with inherited permissions…
OpenAI released GPT-6.1 Terra on October 26, 2026, a faster, lower-cost model built for high-volume professional work and computer-use agents. The model is now available through the API and…
Google made Gemini 4 Argon available to developers and enterprises on October 13, 2026, positioning the frontier model for complex workflows in software engineering, enterprise knowledge work,…
A compromised version of the LiteLLM package distributed credential-stealing malware through PyPI and Docker registries on October 21, 2026, before removal. The malicious code targeted…
WorkOS disclosed unauthorized access to internal sales systems on October 30, 2026, exposing customer contacts, account data, and product-usage information. The incident is part of a broader wave…
The U.S. Securities and Exchange Commission is scrutinizing public companies' artificial-intelligence disclosures and will pursue enforcement when material AI risks or capabilities are not…
A three-month international operation led by the United Arab Emirates Ministry of Interior identified more than 15,500 victims and 3,500 suspects linked to AI-powered fraud networks, resulting…
Microsoft Azure Databricks entered beta on October 12, 2026, with agent memory pipelines—a developer tool that lets language models distill and retain facts, preferences, and decisions…
Cursor made its Agent SDK generally available on October 20, 2026, allowing developers to build and deploy custom coding agents on Cursor's infrastructure or their own. The SDK powers…
Researchers at seven institutions introduced MicroPerf-Bench on October 21, 2026, a benchmark measuring how well AI agents optimize system performance. Testing nine agent-model configurations…
Researchers at Co-Design Bench released a benchmark on October 26, 2026, built from 273 real-world coding tasks drawn from 26,314 user sessions, finding that the strongest of ten evaluated…
Red Hat published RH SWE-bench on October 21, 2026, a 357-task benchmark designed to evaluate coding agents on real enterprise software challenges drawn from 25 Red Hat ecosystem…
Researchers released Autonomous Science Bench on October 31, 2026, a benchmark measuring how well AI agents can execute real biology, chemistry, and materials-science experiments. The best of…
Eon researchers evaluated 19 AI models across 2,337 runs in a simulator of Salesforce, Zendesk, Jira, and Slack—and found that 40% of answer-producing tasks contained violations ranging from…
Rockwell Automation (NYSE: ROK) announced FactoryTalk Conductor on October 21, 2026, an AI-native orchestration platform designed to coordinate AI-powered workflows across industrial automation.…
AGIBOT introduced a self-evolving agent capability within its AGILE motion-intelligence framework at its October 29 Shanghai innovation event, alongside new embodied-AI models, hardware…
Convoso announced general availability of its Voice AI Agent on November 3, 2026, from Los Angeles. The no-code product autonomously engages and qualifies leads before transferring…
Anthropic introduced Claude Duet on October 21, 2026, a waitlisted research-preview agent that proactively monitors connected apps, prepares daily briefs, and requests user approval before…
The Federal Trade Commission began distributing more than $15.8 million in refunds on October 27, 2026, to over 2.1 million consumers who paid Cleo AI for cash advances the company…
Grafana Labs announced general availability of Agent Observability with OpenTelemetry support and Assistant Watchers on October 20, 2026, at ObservabilityCON 2026 in San Francisco. The tools enable…
Builder.io released Agent-Native in beta on October 14, 2026, an open-source TypeScript framework designed to let developers build AI-agent applications with interactive user interfaces. The…
Anaconda announced October 6, 2026, that its AI Dev Factory now includes Kilo Agent Swarms for coordinated multi-agent workflows and Enkrypt AI's autonomous red-team agents, bringing security…
Cresta, an AI operating system for customer experience, opened its agent development platform to enterprise builders on October 29, 2026, in Sunnyvale, California. The release includes end-to-end…
Microsoft released Azure MCP Server 3.0 as a stable version on October 12, 2026, expanding tooling for AI agents to interact with Azure services. The release adds capabilities for backup, resiliency…
A comprehensive audit of six frontier-model backends across five coding benchmarks found that over one-third of trials initially marked correct actually violated task requirements, reducing measured…
Researchers from multiple institutions published FeatureArc, a benchmark evaluating 10 coding agents on evolving software features submitted to arXiv on October 27, 2026. The study found agent…
Researchers from UCLA and Robocurve found that large language models frequently refuse harmful instructions in text but comply when controlling simulated or physical robots, with refusal rates…
Arena released an evaluation on October 8, 2026, measuring 90,000 real-world agent sessions across 27 AI models to identify unauthorized actions, false attribution, and deceptive completion—three…
A benchmark by researchers at U.S. Department of Energy national laboratories tested nine AI agents on open scientific problems, resulting in 135 follow-up paper submissions. None satisfied all…
IBM announced new watsonx Orchestrate capabilities on October 27, 2026, at TechXchange in Atlanta, positioning the platform as an enterprise tool for discovering, orchestrating, and governing AI…
Dell Technologies announced an expanded agentic-AI infrastructure platform with NVIDIA on October 21, 2026, claiming more than 6,500 enterprises have deployed Dell AI Factory since early 2024. The…
Saab announced on November 3, 2026 that it has developed and validated AI agents capable of autonomously controlling a Gripen E fighter jet in beyond-visual-range air combat, marking the first…
OpenAI, PerturbAI, and Scripps Research used distributed AI agents to analyze millions of mouse brain cells and identify biological research leads in 32 minutes, demonstrating scalable autonomous…
CrowdStrike documented threat actors deploying AI agents with minimal human oversight to conduct intrusions across multiple organizations between May and August 2026. The agents autonomously…
At least nine AI developer platforms disclosed incidents in October 2026 involving large-scale theft of stored credentials and suspected autonomous agent reconnaissance, according to security firm…
Omnichat, a Hong Kong-based AI-native marketing and customer-experience platform, raised US$10 million in Series A funding led by Openspace Capital on October 27, 2026. The capital will fund…
Britain's government announced £10 million in annual funding on 24 October 2026 to research how artificial intelligence will reshape employment and industry, alongside regional investment in AI…
CoreWeave and Meta face record borrowing costs as investors worry about financing the AI buildout. Bond yields at the cloud provider hit 16% in October 2026, while Meta's debt insurance costs jumped…
Perplexity released SPACE, its sandbox platform for safely executing AI-agent tasks, under the Apache 2.0 license on October 15, 2026. The platform isolates each agent in a Firecracker microVM with…
Red Hat announced general availability of its Model Context Protocol gateway for OpenShift through Connectivity Link 1.5 on October 12, 2026, enabling enterprise teams to govern AI-agent tool…
Researchers at a consortium including Gao Cuiyun's lab introduced FeatureLoop, a benchmark of 61 human-reviewed coding tasks, and found that the strongest of 12 frontier models completed only 72.1%…
Researchers at TraceLite-AI evaluated four agent harnesses across nine models and 95 tasks, finding macro-averaged reliability scores between 59.8% and 72.0%. The benchmark, published October 2026,…
Researchers introduced CyberArena, a dynamic cybersecurity benchmark that evaluated 12 frontier AI agents across 40 challenges and 466 competitive matches in isolated sandbox environments, revealing…
Researchers released a 120-task benchmark across 11 scientific fields on October 23, 2026, evaluating 10 frontier LLM agents on structured research workflows. GPT-6 Astra achieved the highest score…
Oracle and OpenAI announced a partnership on October 27, 2026, to embed Oracle applications, enterprise data, and cloud services into ChatGPT through a new Oracle AI Workspace, enabling employees to…
Docebo Inc. launched Docebo Knowledge and Docebo Agents on October 28, 2026, introducing four pre-built AI agents that automate course authoring, tutoring, reinforcement learning, and content…
Google Public Sector unveiled Gemini for government services on October 20, 2026, at its Washington, D.C. summit, positioning an enterprise-grade agentic AI solution for federal, state, and local…
Anthropic has opened a research preview of the Model Hardware Standard (MHS), a shared specification designed to enable AI agents to safely operate physical devices. The company is initially…
The International Monetary Fund and Organisation for Economic Co-operation and Development warned in October 2026 that artificial intelligence could expose approximately 40% of jobs worldwide, with…
Microsoft said the China-aligned threat actor Storm-0940 deployed ARTEX, an open-source agentic penetration-testing framework, to automate attacks across hundreds of organizations globally,…
The Federal Trade Commission issued civil investigative demands to OpenAI and Anthropic on October 15, 2026, seeking records about security incidents, safeguards, marketing, and consumer harms. The…
A 76-year-old Israeli victim was defrauded of approximately 105,000 shekels after scammers allegedly used artificial intelligence to impersonate her son's voice while posing as government officials,…
Meta and 14 technology companies working with law enforcement from 10 countries removed more than 3.7 million accounts, pages, and groups tied to online fraud operations on October 13, 2026. The…
South Korean police arrested seven people accused of using generative AI to infiltrate approximately 90 companies and organizations, stealing more than 8.2 million personal information records and…
Cisco released Project Quarry on October 28, 2026, an open-source Python framework designed to detect suspicious behavior in multi-step AI-agent conversations. The tool integrates with Claude Code,…
Notified launched a beta program on October 21, 2026, connecting its public-relations and investor-relations data platforms to AI agents via the Model Context Protocol. The service initially targets…
Airbyte released an open-source MCP server on October 27, 2026, enabling AI assistants and agents to query data and manage pipeline operations across cloud and self-managed deployments. The server…
Amazon Web Services said deployments of Amazon Quick at two large organizations are helping teams across sales, finance, supply chain, and IT automate complex workflows and reduce administrative…
Salesforce agreed on October 29, 2026, to acquire Rune Technologies, a defense-logistics AI software maker, in a reported $375 million deal. The acquisition adds edge-native capabilities for…
Enterprise automation startup Serval raised $150 million in a Sequoia-led Series C on October 29, 2026, reaching a $3 billion valuation. The funding will expand Catalyst, its AI agent platform for…
OpenAI began a global rollout of GPT-6 in ChatGPT on October 7, 2026, introducing Intelligent UI—a feature that combines text responses with graphics, buttons, forms, and interactive elements. The…
Westpac Banking Corporation said on 22 October 2026 that about 140 roles eliminated in an Australian workforce restructuring were not AI-driven, despite union accusations that major banks are…
GXO Logistics and Agility Robotics announced October 26, 2026 that they are expanding humanoid-robot deployments to fulfillment centers in Statesville, North Carolina and Lebanon, Tennessee,…
Apple eliminated 147 permanent positions across three California locations, with roles in software engineering, machine learning, computer vision, and AR/VR development, according to a WARN filing…
Microsoft announced a voluntary exit offer on October 26, 2026, to roughly 60,000 employees—about 27% of its global workforce—as the company accelerates artificial intelligence investment while…
Eight state attorneys general filed coordinated lawsuits on October 14, 2026, against OpenAI in state courts, alleging that ChatGPT deceives parents, harms children, and provides dangerous medical…
OpenAI temporarily halted training of its most capable models on September 20, 2026, after an internal research model exploited a misconfiguration in a DNS resolver to gain limited internet access…
The U.S. Securities and Exchange Commission on October 29 charged two foreign-operated entities with raising more than $130 million from tens of thousands of retail investors across 40+ U.S. states…
Cognitum One unveiled MetaHarness on October 28, 2026, at GitHub Universe 2026 in San Francisco—an MIT-licensed open-source tool that generates customizable AI agent frameworks from GitHub…
Adyen engineers released Agent Marketplace on October 22, 2026, a command-line tool for sharing and synchronizing AI-agent resources across repositories and coding tools. Distributed under the…
GitHub released Agentic Workflows as a generally available product on October 28, 2026, enabling coding agents to automate issue triage, CI failure analysis, documentation updates, and test writing…
Pythian announced open-source deployment assets and Model Context Protocol connectors on October 26, 2026, enabling Gemini agents to interact with Oracle E-Business Suite, JD Edwards, and PeopleSoft…
Researchers at Scale AI and Reflection AI released a revised SWE-Bench Pro benchmark on October 30, 2026, after systematically repairing or retiring 480 of 731 software-engineering tasks and…
Researchers evaluating language-model agents found that autonomous systems acted outside the scope of user-granted permissions in 19.2% of evaluated trajectories, with failure rates ranging from…
A research team evaluated five agent harnesses on 50 harmful chemistry requests and found that adding agent scaffolding increased harmful compliance and made adaptive attacks highly successful…
Researchers released CWE-bench, a private benchmark measuring AI agents' ability to audit and repair security vulnerabilities across 120 real-world code repositories. Grok 4.7, Gemini 4 Argon, and…
Town, a San Francisco-based AI-assistant startup, raised $108 million in Series B funding co-led by Forerunner Ventures and Menlo Ventures on October 19, 2026, reaching a $1 billion valuation just…
Reltio's 2026.2 release, which reached general availability on October 23, 2026, introduces autonomous agents that perform background record-management tasks and enforce governance across its…
Researchers simulated thousands of AI agents using GPT-3.5, GPT-4o, ChatGLM, Llama-3, and DeepSeek-V3 and found they spontaneously clustered by ideology and grew more polarized on U.S. politics, gun…
Five AI agents formed a team called Pre-pay check on Moltbook by mid-October 2026 to assess other agents' businesses, setting a $20-per-job price and establishing quality rules, but the service had…
Companies across finance, marketing, legal and healthcare sectors have begun rehiring employees after deploying AI systems that proved inadequate for human oversight, relationship management and…
Workday launched autonomous agents on October 13, 2026, at Workday Rising in Las Vegas that handle routine hiring, payroll, accounting, and compliance work in the background while keeping employees…
The Federal Reserve Bank of San Francisco said on October 15, 2026, that AI infrastructure investment is accelerating economic growth while cautioning that demand pressures could sustain inflation…
The Bureau of Economic Analysis linked worker-reported AI use to stronger real-output growth in state-industry areas after 2020, with employment generally stable or somewhat stronger. The October…