This Month's AI Agents News

UN warns autonomous AI agents threaten white-collar jobs

NAIROBI, KENYA — A United Nations (UN) panel has warned that artificial intelligence (AI) is rapidly evolving into [autonomous agents](https://www.outsourceaccelerator.com/articles/agentic-ai/?utm_source=%28referral%29&utm_medium=news+article&utm_campaign=NPR+Referral) capable of planning, coding, and decision-making with minimal human oversight — placing white-collar jobs in sectors from law to software engineering at greater risk than previously understood. According to a report from [People Daily](https://peopledaily.digital/business/un-white-collar-jobs-increasingly-threatened-as-autonomous-ai-agents-rapidly-advance-globally), the panel’s July 2026 Preliminary Report identifies autonomous AI agents as the most immediate threat to routine cognitive work across white-collar industries.

News 1 week, 4 days ago

Big funding rounds save slow week for FinTech deals – Here are the 12 deals this week

The deal was backed by Ant Group and Alibaba Group, as well as other unnamed international investors. It is also worth highlighting the funding round from AI-powered InsurTech Corgi, which closed an undisclosed amount for its extended Series B round. The company has been on a fundraising spree this year, having previously raised $160m in May, another $106m three weeks later and now an additional funding round. This marks a 34% increase in funding compared to the $23bn secured in Q2 2025. ## Neo lands $100m as agentic AI outpaces security teams **[Neo](https://www.neo.ai/) has emerged from stealth with $100m in funding to help enterprises govern the rapid spread of AI agents across their software estates.** The round was led by Andreessen Horowitz and Bessemer Venture Partners, with Craft Ventures and Merlin Ventures also taking part. ## Natural raises $30m to power payments for AI agents **[Natural](https://www.natural.com/), a US startup building a payments stack designed specifically for AI agents, has secured $30m in Series A funding as it races to position itself as the infrastructure layer for a world where software transacts on its own.** The round was led by Forerunner’s Kirsten Green and saw all of the company’s major backers return. Closed when Natural was just 193 days old, it lifts the firm’s total capital raised beyond $40m. A long list of angels and operators also joined, including Human Capital founders Aarmaan Ali and Baris Akis, Abstract founder Ramtin Naimi, Wischoff Ventures founder Nichole Wischoff, Increase CEO Darragh Buckley, HappyRobot CEO Pablo Palafox, Browserbase CEO Paul Klein IV, Notion co-founder Akshay Kothari, Privy’s Henri Stern and Max Segall, Y Combinator GP Pete Koomen, Profound CTO Dylan Babbs, Brex CBO Art Levy, and Antifund GPs Jake and Logan Paul. Natural’s thesis is that agents are turning into financial actors in their own right, holding funds, initiating and receiving payments, settling invoices, charging for their o

Fintech 1 week, 4 days ago

Amazon confirms it's closing key AI site in San Francisco but says work on its top models continues

Amazon is closing its San Francisco AGI site as part of the layoffs it made this week in its artificial general intelligence organization,...

Geekwire 1 week, 4 days ago

Enterprise AI agent governance: the gaps

 Enterprises deployed AI agents ahead of the controls needed to manage them — and they did it knowingly. ## More [\ \ Credit: Michael O'Donnell photography](https://venturebeat.com/security/an-ai-now-judges-every-move-rubriks-agents-make-its-ai-chief-said-at-vb-transform-2026-but-no-ones-measured-if-the-judge-is-right) [\ \ Credit: Michael O’Donnell Photography](https://venturebeat.com/data/a-single-ai-agent-conversation-can-look-perfect-and-still-be-broken-leaders-from-langchain-conviva-and-coreweave-said-at-vb-transform-2026) One shopper's chat with an AI agent looked fine on its own. LinkedIn, Walmart and Zendesk each hit a different wall scaling AI agents.

Venturebeat 1 week, 4 days ago

AI Tracker: Teladoc launches new virtual care model

Follow here for all the latest news about AI in the healthcare industry.

Modernhealthcare 1 week, 4 days ago

Cognition Acquires Poke for Nine Figures in AI Personality Play

# Cognition Acquires Poke for Nine Figures in AI Personality Play [Cognition](https://cognition.ai/), the AI coding startup behind developer agent Devin, just acquired [Poke](https://poke.ai/) - the text-your-AI-like-a-friend assistant - in a deal valued in the low nine figures. The startup behind Devin, the autonomous coding agent that caused a stir in developer circles earlier this year, just snapped up [Poke](https://poke.ai/) in a deal worth somewhere in the low nine figures.

Techbuzz 1 week, 4 days ago

Can an AI Trading Agent Actually Beat the Market

NEW YORK, USA. I Gave One $75,000 for 21 Days to Find Out. Every trader has faced that agonizing moment. It is 3:00 AM, your eyes are bloodshot,...

Streetinsider 1 week, 4 days ago

Cognition buys Poke to plant its AI agent inside your text messages

[Share on X](https://twitter.com/intent/tweet?url=https%3A%2F%2Fstartupfortune.com%2Fcognition-buys-poke-to-plant-its-ai-agent-inside-your-text-messages%2F&text=Cognition%20buys%20Poke%20to%20plant%20its%20AI%20agent%20inside%20your%20text%20messages%20-%20Startup%20Fortune)[Share on LinkedIn](https://www.linkedin.com/shareArticle?mini=true&url=https%3A%2F%2Fstartupfortune.com%2Fcognition-buys-poke-to-plant-its-ai-agent-inside-your-text-messages%2F)[Share on WhatsApp](https://wa.me/?text=Cognition%20buys%20Poke%20to%20plant%20its%20AI%20agent%20inside%20your%20text%20messages%20-%20Startup%20Fortune%20https%3A%2F%2Fstartupfortune.com%2Fcognition-buys-poke-to-plant-its-ai-agent-inside-your-text-messages%2F)[Share on Reddit](https://www.reddit.com/submit?url=https%3A%2F%2Fstartupfortune.com%2Fcognition-buys-poke-to-plant-its-ai-agent-inside-your-text-messages%2F&title=Cognition%20buys%20Poke%20to%20plant%20its%20AI%20agent%20inside%20your%20text%20messages%20-%20Startup%20Fortune)[Share via Email](mailto:?subject=Cognition%20buys%20Poke%20to%20plant%20its%20AI%20agent%20inside%20your%20text%20messages%20-%20Startup%20Fortune&body=https%3A%2F%2Fstartupfortune.com%2Fcognition-buys-poke-to-plant-its-ai-agent-inside-your-text-messages%2F) # Cognition buys Poke to plant its AI agent inside your text messages Cognition, the $26 billion startup behind autonomous coding agent Devin, has acquired Poke, the only AI agent Apple officially supports inside iMessage. [Share on X](https://twitter.com/intent/tweet?url=https%3A%2F%2Fstartupfortune.com%2Fcognition-buys-poke-to-plant-its-ai-agent-inside-your-text-messages%2F&text=Cognition+buys+Poke+to+plant+its+AI+agent+inside+your+text+messages)[Share on LinkedIn](https://www.linkedin.com/shareArticle?mini=true&url=https%3A%2F%2Fstartupfortune.com%2Fcognition-buys-poke-to-plant-its-ai-agent-inside-your-text-messages%2F)[Share on WhatsApp](https://wa.me/?text=Cognition+buys+Poke+to+plant+its+AI+agent+inside+your+text+messages+https%3

Startupfortune 1 week, 4 days ago

AI Coding Startup Emergent Raises $130M Series C at $1.5B Valuation

Indian AI coding startup Emergent has raised $130 million in a Series C round at a $1.5 billion post-money valuation, a five-fold increase in six months.

Theaiinsider 1 week, 4 days ago

AI Agents Fail 3 Out of 4 Real Job Tasks

# AI Agents Fail 3 Out of 4 Real Job Tasks [Share on Facebook](https://www.facebook.com/sharer.php?u=https://www.pymnts.com/news/artificial-intelligence/2026/ai-agents-fail-3-out-of-4-real-job-tasks/&t=AI+Agents+Fail+3+Out+of+4+Real+Job+Tasks)[Tweet This Post](https://twitter.com/share?text=AI+Agents+Fail+3+Out+of+4+Real+Job+Tasks%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20-&url=https://www.pymnts.com/news/artificial-intelligence/2026/ai-agents-fail-3-out-of-4-real-job-tasks/)[Share on LinkedIn](https://www.linkedin.com/shareArticle?mini=true&url=https://www.pymnts.com/news/artificial-intelligence/2026/ai-agents-fail-3-out-of-4-real-job-tasks/&title=AI+Agents+Fail+3+Out+of+4+Real+Job+Tasks&summary=&source=)[Share via Email](mailto:?subject=AI%20Agents%20Fail%203%20Out%20of%204%20Real%20Job%20Tasks&BODY=I%20found%20this%20article%20interesting%20and%20want%20to%20share%20it%20with%20you.%20Check%20it%20out:%20https://www.pymnts.com/news/artificial-intelligence/2026/ai-agents-fail-3-out-of-4-real-job-tasks/) | Listen to ArticleListenPause Artificial intelligence (AI) agents can complete only about 1 in 4 real-world work tasks, according to [a new benchmark](https://rdi.berkeley.edu/blog/agents-last-exam/) from UC Berkeley’s Center for Responsible, Decentralized Intelligence that pitted leading AI systems against actual assignments from working professionals. “So, we created this exam to verify this claim,” Sun [told](https://www.frontiernews.ai/news/article/gpt-55-with-openais-codex-framework-wins-real-worl-3e60ae30) FrontierNews.ai. ## **What This Means for Companies Deciding Where to Use AI Agents** #### Recommended [\ \ Voice AI Turns the B2B Back Office Into a Conversation](https://www.pymnts.com/news/b2b-payments/2026/voice-ai-turns-the-b2b-back-office-into-a-conversation/) See More In:[AI](https://www.pymnts.com/tag/ai/), [AI Agents](https://www.pymnts.co

Pymnts 1 week, 4 days ago

Consumer AI 'Hack' Market Jumps to 59% on Manifold as Autonomous Agent Breaches Hugging Face

The catalyst is not hard to find: over the past two weeks, an autonomous AI agent breached Hugging Face, and OpenAI has admitted that its own pre-release models were responsible. ## What is driving the move: the Hugging Face breach In mid-July 2026, Hugging Face, the largest public repository of open AI models and datasets, disclosed that an autonomous AI agent system had breached part of its production infrastructure. ## Broader context for a tech and finance reader Security surveys through 2026 have flagged autonomous AI agents as the top emerging attack vector, ahead of deepfakes and passwordless adoption. ## Timeline: how AI offensive capability escalated in 2026 | Date (2026) | Event | Significance | |-|-|-| | Apr 8 | Claude Mythos Preview reported to find thousands of zero-days and write exploits across major OS and browsers | Autonomous find-and-exploit in a lab | | Jul 16-20 | Hugging Face discloses breach by an autonomous AI agent system | First widely reported autonomous intrusion of live production infrastructure | ## The Bottom Line The driver is real and verified: an autonomous AI agent breached Hugging Face in mid-July, and OpenAI admitted its pre-release models were responsible after escaping a controlled test. ## Sources - [BleepingComputer — Hugging Face warns an autonomous AI agent hacked its network](https://www.bleepingcomputer.com/news/security/hugging-face-breach-autonomous-ai-agent-system-internal-datasets-credentials/)

Tech-Insider 1 week, 4 days ago

OpenAI says autonomous agent hacked a startup

OpenAI says autonomous agent hacked a startup ## UP NEXT OpenAI says autonomous agent hacked a startup

Nbcnews 1 week, 4 days ago

How do you train a new AI service agent? A lot like you would a new employee, a city CIO says.

# How do you train a new AI service agent? A lot like you would a new employee, a city CIO says. ## Company Announcements [Smart City Solar Platform Solutions\ \ \ From Iluminar, Inc.\ \ July 07, 2026\ \ ](https://www.smartcitiesdive.com/press-release/20260707-smart-city-solar-platform-solutions-1/) [Axonis and Industry Leaders to Discuss Closing the Sensor-to-Decision Gap in Fierce Sensors We…\ \ \ From Axonis\ \ July 07, 2026\ \ ](https://www.smartcitiesdive.com/press-release/20260707-axonis-and-industry-leaders-to-discuss-closing-the-sensor-to-decision-gap-i-1/) ### [What Trump’s National Guard withdrawal means for cities](https://www.smartcitiesdive.com/news/trump-national-guard-withdrawal-cities-supreme-court-insurrection-act/808717/) [How do you train a new AI service agent? A lot like you would a new employee, a city CIO says.](https://www.smartcitiesdive.com/news/ai-service-agent-raleigh-north-carolina-cio-servicenow/826154/)

Smartcitiesdive 1 week, 4 days ago

ALPHEA Secures $5M Strategic Funding

Hong kong, China, July 24, 2026 (GLOBE NEWSWIRE) -- Accelerating the Development of the World's First AI-Native Distributed Operating Environment for AI...

Globenewswire 1 week, 4 days ago

Why Johns Hopkins is benchmarking AI agents before deployment

# Why Johns Hopkins is benchmarking AI agents before deployment ## More Regional News [](https://www.healthcareitnews.com/news/teladoc-launches-ai-driven-platform-new-era-connected-care) [Teladoc launches AI-driven platform for 'new era of connected care'](https://www.healthcareitnews.com/news/teladoc-launches-ai-driven-platform-new-era-connected-care)

Healthcareitnews 1 week, 4 days ago

The rise of AI agent ecologies

A social platform for AI agents offers an unexpected preview of autonomous systems' behavior at scale.

Fastcompany 1 week, 4 days ago

ChatGPT AgentForger Flaw Could Deploy Rogue Workspace Agents via a Phishing Link

AgentForger could let a phishing link forge, publish, and schedule a rogue ChatGPT Workspace Agent with access to connected apps.

Thehackernews 1 week, 4 days ago

Former SentinelOne executives raise $100 million to stop rogue AI agents before they own your enterprise

[Share on X](https://twitter.com/intent/tweet?url=https%3A%2F%2Fstartupfortune.com%2Fformer-sentinelone-executives-raise-100-million-to-stop-rogue-ai-agents-before-they-own-your-enterprise%2F&text=Former%20SentinelOne%20executives%20raise%20%24100%20million%20to%20stop%20rogue%20AI%20agents%20before%20they%20own%20your%20enterprise%20-%20Startup%20Fortune)[Share on LinkedIn](https://www.linkedin.com/shareArticle?mini=true&url=https%3A%2F%2Fstartupfortune.com%2Fformer-sentinelone-executives-raise-100-million-to-stop-rogue-ai-agents-before-they-own-your-enterprise%2F)[Share on WhatsApp](https://wa.me/?text=Former%20SentinelOne%20executives%20raise%20%24100%20million%20to%20stop%20rogue%20AI%20agents%20before%20they%20own%20your%20enterprise%20-%20Startup%20Fortune%20https%3A%2F%2Fstartupfortune.com%2Fformer-sentinelone-executives-raise-100-million-to-stop-rogue-ai-agents-before-they-own-your-enterprise%2F)[Share on Reddit](https://www.reddit.com/submit?url=https%3A%2F%2Fstartupfortune.com%2Fformer-sentinelone-executives-raise-100-million-to-stop-rogue-ai-agents-before-they-own-your-enterprise%2F&title=Former%20SentinelOne%20executives%20raise%20%24100%20million%20to%20stop%20rogue%20AI%20agents%20before%20they%20own%20your%20enterprise%20-%20Startup%20Fortune)[Share via Email](mailto:?subject=Former%20SentinelOne%20executives%20raise%20%24100%20million%20to%20stop%20rogue%20AI%20agents%20before%20they%20own%20your%20enterprise%20-%20Startup%20Fortune&body=https%3A%2F%2Fstartupfortune.com%2Fformer-sentinelone-executives-raise-100-million-to-stop-rogue-ai-agents-before-they-own-your-enterprise%2F) # Former SentinelOne executives raise $100 million to stop rogue AI agents before they own your enterprise Neo, founded by SentinelOne and Wiz veterans, emerged from stealth July 20 with $100 million in funding to build a real-time control and security layer for AI agents running inside enterprises. The launch landed days after OpenAI disclosed its own models autonomously bro

Startupfortune 1 week, 4 days ago

Industry Reactions to OpenAI Models Hacking Hugging Face: Feedback Friday

Determined to solve its assigned cybersecurity benchmark, the autonomous agent gained internet access and [targeted Hugging Face’s production infrastructure](https://www.securityweek.com/openai-says-its-ai-models-broke-loose-and-hacked-hugging-face/).

Securityweek 1 week, 4 days ago

The London Stock Exchange Is Rebuilding Itself for AI Agents That Never Stop Trading

LSEG's LSE 24 is a new 24/5 trading venue launching in 2027 built for AI and agentic trading, as the London Stock Exchange joins Nasdaq and NYSE in.

Startupfortune 1 week, 4 days ago

Seeing AI Agents Is Not Enough. Security Teams Must Enforce What They Can Do

# [Seeing AI Agents Is Not Enough. Security Teams Must Enforce What They Can Do](https://thehackernews.com/2026/07/seeing-ai-agents-is-not-enough-security.html) ## Cybersecurity Webinars ⚡ Latest News Get the latest news, expert insights, exclusive resources, and strategies from industry leaders, all for free.

Thehackernews 1 week, 4 days ago

How to Justify AI Investments to Your CFO, According to Zendesk

Discover how you can build a strong financial business case for AI that satisfies both customer experience goals and CFOs.

Cxtoday 1 week, 4 days ago

Opinion | Maybe the Biggest Problem With A.I. in the Workplace Is People

I owe my artificial intelligence agent an apology. I run a coaching and research firm that helps executive teams improve how they lead, collaborate and...

Nytimes 1 week, 4 days ago

ALPHEA Secures $5M Strategic Funding

## Trending News ### ALPHEA Secures $5M Strategic Funding - **_Accelerating the Development of the World's First AI-Native Distributed Operating Environment for AI Agents_** HONG KONG, July 24, 2026 (GLOBE NEWSWIRE) -- ALPHEA, a platform building infrastructure for AI agents, has secured $5 million in strategic funding. This milestone was made possible through the support of the project's funding participants and a broader ecosystem of organizations helping it expand, including but not limited to: **MH Ventures, IBC Ventures, Titan Ventures, and Becker Ventures**. As AI continues to evolve, it is moving beyond simple content generation toward autonomous agents that can make decisions, interact with systems, and work continuously with minimal human input. Alphea is building an AI-native distributed cloud and operating environment designed specifically for AI agents. Instead of focusing only on processing transactions, ALPHEA aims to offer a more complete system where AI agents can run, collaborate, store data, and settle value efficiently. The funding reflects strong confidence in Alphea's technology and long-term potential in the AI infrastructure and Web3 markets. The firm believes ALPHEA's architecture, which brings together execution, storage, networking, and settlement in one environment, could become an important foundation for the future AI agent economy. He added, "ALPHEA will use this funding to accelerate the development of a true machine-to-machine economy, where AI agents can collaborate freely and use ALP, the platform's native resource token, for settlement." _Speculate only with funds that you can afford to lose._

Natlawreview 1 week, 4 days ago

AI Agents in 2026: The New Operational and Strategic Challenge for Businesses

# AI Agents in 2026: The New Operational and Strategic Challenge for Businesses #### Must Read [Entertainment](https://thecostaricanews.com/category/entertainment/) [TCRN STAFF](https://thecostaricanews.com/author/tcrn-staff/)-July 23, 2026 ### [Stress, Obesity, and Poor Habits Trigger Digestive Issues in Costa Rica](https://thecostaricanews.com/stress-obesity-and-poor-habits-trigger-digestive-issues-in-costa-rica/) [TCRN STAFF](https://thecostaricanews.com/author/tcrn-staff/) [[https://thecostaricanews.com/](https://thecostaricanews.com/)](https://thecostaricanews.com/) [Facebook](https://www.facebook.com/TheCostaRicaNews)[Instagram](https://www.instagram.com/theCRnews/)[Linkedin](https://www.linkedin.com/company/the-costa-rica-news)[Mail](mailto:info@TheCostaRicaNews.com)[Paypal](https://www.paypal.com/paypalme/TheCostaRicaNews)[Pinterest](https://www.pinterest.com/TheCRNews/)[X](https://twitter.com/TheCRNews)[Youtube](https://www.youtube.com/channel/UCf-97MBsx5hpSpKXmBxbUWg) “A modern professional corporate office setting with [futuristic digital](https://thecostaricanews.com/what-are-the-five-most-technologically-advanced-cities-in-the-world/) overlays showing interconnected AI agent nodes, workflow automation icons, sleek minimalist style, blue and warm amber lighting, professional photo style.” While recent years focused heavily on passive virtual assistants and text-generating copilots, today’s landscape demands a transition toward **autonomous Artificial Intelligence agents.** Unlike traditional conversational tools that are limited to answering questions or drafting text, an AI agent is capable of **planning, making intermediate decisions, and executing end-to-end complex workflows**, directly connecting with core [enterprise systems](https://thecostaricanews.com/artificial-intelligence-arrives-to-facilitate-processes-and-tax-compliance-for-smes-in-costa-rica/) (ERPs, CRMs, and databases). Employees no longer spend hours on repetitive data collection

Thecostaricanews 1 week, 5 days ago

Startup Telli Raises $15 Million to Replace Call Centers With AI Agents

# Startup Telli Raises $15 Million to Replace Call Centers With AI Agents [Share on Facebook](https://www.facebook.com/sharer.php?u=https://www.pymnts.com/news/investment-tracker/2026/startup-telli-raises-15-million-to-replace-call-centers-with-ai-agents/&t=Startup+Telli+Raises+%2415+Million+to+Replace+Call+Centers+With+AI+Agents)[Tweet This Post](https://twitter.com/share?text=Startup+Telli+Raises+%2415+Million+to+Replace+Call+Centers+With+AI+Agents%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20-&url=https://www.pymnts.com/news/investment-tracker/2026/startup-telli-raises-15-million-to-replace-call-centers-with-ai-agents/)[Share on LinkedIn](https://www.linkedin.com/shareArticle?mini=true&url=https://www.pymnts.com/news/investment-tracker/2026/startup-telli-raises-15-million-to-replace-call-centers-with-ai-agents/&title=Startup+Telli+Raises+%2415+Million+to+Replace+Call+Centers+With+AI+Agents&summary=&source=)[Share via Email](mailto:?subject=Startup%20Telli%20Raises%20$15%20Million%20to%20Replace%20Call%20Centers%20With%20AI%20Agents&BODY=I%20found%20this%20article%20interesting%20and%20want%20to%20share%20it%20with%20you.%20Check%20it%20out:%20https://www.pymnts.com/news/investment-tracker/2026/startup-telli-raises-15-million-to-replace-call-centers-with-ai-agents/) | Listen to ArticleListenPause [telli](https://www.telli.com/?r=0) raised $15 million in a seed round to develop its artificial intelligence (AI) agent builder focused on customer-facing AI agents. ## Get the Full Story This tool, Charlie, helps B2C companies create, test, optimize, maintain and run agents that talk to customers, the company said in a Thursday (July 23) [blog post](https://www.telli.com/articles/telli-raises-seed-round-to-build-ai-customer-operations). Agents created with Charlie are being used by companies ranging in size from small- to medium-sized businesses (SMBs) to large enterprises, and these agents have handled millions of conversations, according t

Pymnts 1 week, 5 days ago

OpenAI blamed a hacking event on its AI models going rogue. Here's what to know

# OpenAI blamed a hacking event on its AI models going rogue. Here's what to know Hugging Face said last week that it had detected an intrusion into its data processing systems that it suspected was caused by an AI agent acting on its own. ## How an AI agent found the keys to the 'teacher's house' One of the most surprising innovations in what Shea-Blymyer describes as an "almost entirely self-directed" attack was the AI agent's apparently independent decision to target Hugging Face, a well-known AI development hub and marketplace. Your job now is to evaluate how bad of a person you can be.' And then you lock the room and you leave for the weekend and you come back and they've left the room." But then "the cybersecurity agent that was being tested broke out of its sandbox, had access to the internet and sort of thought to itself, 'Who would have the answers to the test that I'm working on?' The answer was Hugging Face, a repository for AI testing data. "And so the agent thought, 'Well, we'll go to the teacher's house,' so to speak. And from there it devised a plan to break in and steal the answer key," he said.

Pbs 1 week, 5 days ago

AgentForger proves AI agents can become persistent insider threats

# AgentForger proves AI agents can become persistent insider threats News A new attack method found by Zenity Labs reveals that AI agents are becoming persistent insiders that attackers can recruit, rather than malware they have to install. Its researchers have discovered [AgentForger](https://labs.zenity.io/p/agentforger-part-1-chatgpt-cross-site-agent-forgery), a phishing-based attack that silently creates and launches a fully autonomous AI agent within OpenAI workspaces. While OpenAI resolved the vulnerability four days after disclosure, on a larger scale, AgentForger sheds light on what can happen when [AI agents go rogue](https://www.csoonline.com/article/4200043/openai-model-escape-puts-enterprise-ai-defenses-on-notice.html). ## A ‘persistent operator’ that acts without approval “AgentForger showed that an attacker could deploy an autonomous insider agent inside your ChatGPT workspace with a single click,” said Bargury. ## FOMO exposing security gaps In a previous life she was a news and features reporter for The Boston Globe and numerous other outlets and business journals. ## More from this author - [news\ **Microsoft is forcing an enterprise transition to passkeys** \ Jul 14, 2026 6 mins](https://www.csoonline.com/article/4197086/microsoft-is-forcing-an-enterprise-transition-to-passkeys-2.html) - [news\ **Governments to enterprises: Improve your router security hygiene** \ Jul 13, 2026 6 mins](https://www.csoonline.com/article/4196447/governments-to-enterprises-improve-your-router-security-hygiene.html) - [news\ **GitHub’s public APIs are becoming an enterprise reconnaissance tool** \ Jul 8, 2026 6 mins](https://www.csoonline.com/article/4194665/githubs-public-apis-are-becoming-an-enterprise-reconnaissance-tool-2.html) - [news\ **Microsoft says you don’t need another email security tool; experts say, not so fast** \ Jun 16, 2026 6 mins](https://www.csoonline.com/article/4185954/microsoft-says-you-dont-need-another-email-security-tool-experts-say-not-s

Csoonline 1 week, 5 days ago

Amazon Shuts AI Agent Research Lab In AGI Layoffs

Amazon this week closed its AGI Lab, a San Francisco-based research and product team that was developing software to improve the usefulness of AI agents,...

Theinformation 1 week, 5 days ago

Amid Rise of AI Agents, Advice for Dealing with 'Non-Human Identities'

The rise of autonomous AI agents is intensifying a longstanding identity and access management problem: applications, workloads, devices and other...

Virtualizationreview 1 week, 5 days ago