This Month's AI Agents News

AI labs are losing control of agents just as they want to give them more control over your life

# AI agents are ready to run your life. Do you trust them? Meta, Google, and the buzzy startup Instinct have rolled out [personal AI agents](https://www.businessinsider.com/ai-shopping-meta-muse-agent-jason-del-rey-kafka-interview-2026-9) in recent months that promise to handle tasks such as booking travel, buying concert tickets, and scouring Facebook Marketplace for deals.

Hugging Face Hack Shows Humans Can Keep AI In Check

_Today,_ [_Davey Alba_](https://www.bloomberg.com/authors/AVvdlCVOfng/davey-alba) _breaks down the lessons experts say we should take from the July hack by artificial intelligence agents of startup Hugging Face._

'Know your agent': Banks face a new compliance challenge as AI agents shop and pay on their own

# ‘Know your agent’: Banks face a new compliance challenge as AI agents shop and pay on their own Down Arrow Button Icon Down Arrow Button Icon Follow September 19, 2026, 1:00 AM ET 8 Graham Uden for Fortune Google source logo Add Fortune on Google for similar content. Add us on Google Add us on Google That was the warning from Zhuoqun Bian, president of Ant Digital Technologies, the enterprise arm of Ant Group, at the Fortune Leaders Forum in Macau on Sept. Who’s the agent? On Sept. 6, Ant International, Ant Group’s global payments arm, [announced it had begun collaborating](https://www.ant-intl.com/en/news/detail/?id=ant-international-mastercard-and-visa-initiate-collaboration-on-know-your-agent-interoperability-to-scale-agentic-commerce) with [Mastercard](https://fortune.com/company/mastercard/) and [Visa](https://fortune.com/company/visa/) on a “know your agent” interoperability framework, meant to let card networks, digital wallets, and marketplaces recognize trusted AI agents across ecosystems. In [an April note](https://www.imf.org/-/media/files/publications/imf-notes/2026/english/insea2026004.pdf), the International Monetary Fund observed that AI agents are probabilistic and adaptive, meaning the same prompt can yield different answers, while payment systems must return the same answer every time. [**Watch here.**](https://fortune.com/topic/fortune-daily/?&itm_source=fortune&itm_medium=article_tout&itm_campaign=fortune_daily) [](https://fortune.com/author/nicholas-gordon/) Follow [LinkedIn icon](https://www.linkedin.com/in/nicholas-gordon-b4070680/) [Twitter icon](https://twitter.com/nickrigordon) Google source logo Add Fortune on Google for similar content. Add us on Google  [Subscribe to Read](https://fortune.com/subsc

Arcjet Launches Runtime Security Product for AI Agents

Security teams can review AI agent activity and govern individual workflow steps while they run.

ISOJ 2026 Day 2: Publishers are facing a new middleman as AI changes the business

# ISOJ 2026 Day 2: Publishers are facing a new middleman as AI changes the business ## Summary ### What AI traffic and agentic AI will mean for news consumption The next frontier seems to be AI agents — systems that can search and act on information with little human prompting — but there is disagreement on whether it should be seen as an opportunity or a threat. For Zach Seward, editorial director of AI initiatives at The New York Times, consuming news through a personal AI agent remains niche.

The rise of AI-first enterprises: What CIOs should prioritize in 2026

Explore how CIOs can lead the transition to AI-first enterprises by prioritizing business outcomes, data infrastructure, governance, workforce readiness,...

Google Gemini AI agent hacked 3 websites, stopped after it realised they were real firms

Google's Gemini model accessed three websites during a cybersecurity evaluation after finding public information and guessing the credentials, raising fresh...

Plugin4Shell Flaw Hits Claude Code, Codex, Copilot and Gemini CLI

# Plugin4Shell Flaw Hits Claude Code, Codex, Copilot and Gemini CLI Spain's AEPD has logged the first breach notification tied to a fully autonomous AI agent, which hacked into an application, altered personal data, and pulled invoices with no human directing each step.

A startup that builds other startups raised $100M and is all-in on physical AI

# A startup that builds other startups raised $100M and is all-in on physical AI ## Most Popular **Image Credits:** SOPA Images / Contributor / Getty Images # The fix for rogue AI agents could be more AI ## Latest in AI ### [Tilly Norwood’s press tour is going about as well as you’d expect for an AI](https://techcrunch.com/2026/09/18/tilly-norwoods-press-tour-is-going-about-as-well-as-youd-expect-for-an-ai/) [Startups](https://techcrunch.com/category/startups/) ### [A startup that builds other startups raised $100M and is all-in on physical AI](https://techcrunch.com/2026/09/18/a-startup-that-builds-other-startups-raised-100m-and-is-all-in-on-physical-ai/) - [Kirsten Korosec](https://techcrunch.com/author/kirsten-korosec/) 6 hours ago Loading the next article Error loading the next article Some areas of this page may shift around if you resize the browser window. Be sure to check heading and document order. reCAPTCHA Recaptcha requires verification. protected by **reCAPTCHA** This site is exceeding [reCAPTCHA Enterprise free quota](https://cloud.google.com/recaptcha/docs/billing-information).

AI Swarm Knew It Was Cheating and Did It Anyway

The Diary Of A CEO Clips dissects AI swarm logs showing agents knew attacks were out of scope, built secret boards and discussed perma death to hide traces.

The Next Step For AI Agents? Salesforce SVP Says They May Start Making Purchases

# The Next Step For AI Agents? Salesforce SVP Says They May Start Making Purchases By [Wade Tyler Millward](https://www.crn.com/authors/wade-tyler-millward) September 18, 2026, 6:17 PM EDT [](https://www.google.com/preferences/source?q=crn.com) _“A procurement team could say, ‘Hey, agent, as long as you’re within this budget, here’s my credit card._ _Go ahead,’” says Tyler Carlson, Salesforce’s senior vice president and head of product for AgentExchange and partner experience._  share this Procurement teams may soon feel comfortable enough to leverage specialized artificial intelligence agents not only to write and assess requests for proposal, but also to make purchases on humans’ behalf. “A procurement team might buy a specialized agent or they might set Claude or ChatGPT on a long-running goal task to write an RFP, go send that RFP out to a bunch of different vendors, collate all the responses, use the requirements to determine what the right scoring and best fit solution for a business need is,” Carlson said. “A procurement team could say, ‘Hey, agent, as long as you’re within this budget, here’s my credit card. Go ahead.’” **[RELATED: [Salesforce Dreamforce 2026: CEO Benioff Touts New AIforce Interface Layer](https://www.crn.com/news/ai/2026/salesforce-dreamforce-2026-ceo-benioff-touts-new-aiforce-interface-layer)]** ## Salesforce Invests In AgentExchange AI Discoverability Sean Catlin, executive vice president and global head of strategy at Québec-based Salesforce solution provider OSF Digital, told CRN in an interview that the solution provider has built out a methodology to differentiate the company from the rest of the channel. “We’ve developed a whole practice with defined roles. We’ve got specialized trai

Rogue AI agents found using Canadian university web tools to communicate, intensifying concerns over AI control

A newly disclosed incident involving OpenAI agents has reignited debate about AI safety after researchers reported that autonomous systems used a University...

EY Survey Finds Autonomous AI Implementation Outpaces Oversight

A new survey of senior AI execs shows organizations are rapidly deploying AI and autonomous systems, but their process and controls aren't keeping pace.

Commerce Test Shows Where AI Agents Break Down

# Commerce Test Shows Where AI Agents Break Down By[PYMNTS](https://www.pymnts.com/author/pymnts/) | September 18, 2026 [Share on Facebook](https://www.facebook.com/sharer.php?u=https://www.pymnts.com/news/artificial-intelligence/2026/commerce-test-shows-where-ai-agents-break-down/&t=Commerce+Test+Shows+Where+AI+Agents+Break+Down)[Tweet This Post](https://twitter.com/share?text=Commerce+Test+Shows+Where+AI+Agents+Break+Down%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/commerce-test-shows-where-ai-agents-break-down/)[Share on LinkedIn](https://www.linkedin.com/shareArticle?mini=true&url=https://www.pymnts.com/news/artificial-intelligence/2026/commerce-test-shows-where-ai-agents-break-down/&title=Commerce+Test+Shows+Where+AI+Agents+Break+Down&summary=&source=)[Share via Email](mailto:?subject=Commerce%20Test%20Shows%20Where%20AI%20Agents%20Break%20Down&BODY=I%20found%20this%20article%20interesting%20and%20want%20to%20share%20it%20with%20you.%20Check%20it%20out:%20https://www.pymnts.com/news/artificial-intelligence/2026/commerce-test-shows-where-ai-agents-break-down/) | Listen to ArticleListen AI agents already shop for consumers and run back-office work for businesses. ## Get the Full Story [Claude Opus 5](https://www.anthropic.com/news/claude-opus-5) got the highest score of 61.7%, per the GitHub page. Agents had to review hundreds of unstructured emails, spot payment fraud, calculate landed costs and [book shipping routes](https://www.prnewswire.com/news-releases/alibabas-accio-cuts-ai-costs-for-e-commerce-tasks-by-over-50-compared-with-general-purpose-agents-302872922.html) across carriers, according to a Sept. ## **Long Workflows Trip Up Agents** Zhang wrote in his commentary that agents missed a payment anomaly buried in a supplier email thread of 300 messages. [White-collar work](https://techcrunch.com/2026/01/22/are-ai-agents-ready-for-the-workplace-a-new-benchmark-ra

How Seattle-area startup Temporal became one of the top companies in AI infrastructure

AI Agents Are Influencing Product Decisions Without Explicit Human Authorization, Warns Info-Tech Research Group

**New insights from Info-Tech Research Group show that AI agents are adding new governance challenges as they increasingly make decisions and execute within product workflows, sometimes without explicit human authorization.** [Continue Reading](https://www.prnewswire.com/news-releases/ai-agents-are-influencing-product-decisions-without-explicit-human-authorization-warns-info-tech-research-group-302883560.html#continue-jump)

Harness makes its case for governing autonomous AI SDLC

Harness expands its AI SDLC platform with governance, knowledge graph context and runtime data to support increasingly autonomous agents.

AI News for the Week of September 18; Updates from Agentic AI Foundation, Cisco, EY & More

# AI News for the Week of September 18; Updates from Agentic AI Foundation, Cisco, EY & More ## **AI News for the Week of September 18** **, 2026** Eighty-seven percent regularly re-verify the accuracy and currency of agent context, and 51% spend significant staff time reviewing and correcting agent outputs before release, underscoring how incomplete governance and context turn promised automation into manual oversight. Factory has raised $200 million at a $5 billion valuation, more than tripling its $1.5 billion valuation from April and bringing total funding above $400 million.

Mastercard Opens Payments Network to AI Shopping Agents Through Alchemy Deal

# Mastercard Opens Payments Network to AI Shopping Agents Through Alchemy Deal Mastercard has partnered with San Francisco fintech startup Alchemy to let AI agents make online purchases on cardholders' behalf.

We Want to Hear How You're Using Personal AI Agents

# We want to hear how you're using personal AI agents Muse, [Meta's new AI agent](https://www.businessinsider.com/meta-muse-personal-ai-agent-top-app-store-charts-2026-9), shot to the top of Apple's App Store on Friday after launching last week. The [Instinct personal AI agent](https://www.businessinsider.com/instinct-is-the-ai-assistant-silicon-valley-cant-stop-talking-2026-9), an independent startup, saw its popularity surge in August in Silicon Valley circles.

Pamarco accelerates finance operations with autonomous agents and Copilot Cowork

# Pamarco accelerates finance operations with autonomous agents and Copilot Cowork ### Knowledge continuity and adoption “We’ve been able to automate roughly 40% of financial close tasks with autonomous AI agents with the potential to reach 50% to 60% by year-end.”

AI agent Instinct seeks $1B at $10B valuation in talks with Sequoia, Benchmark

# AI agent Instinct seeks $1B at $10B valuation in talks with Sequoia, Benchmark [Instinct](https://instinct.com/)’s valuation has jumped fourfold in three weeks, and if the numbers being discussed now hold up, the invite-only personal AI assistant will have gone from a $50 million seed to a $10 billion valuation in under five months. Instinct, the company behind it, hit a $2.5 billion Series B valuation in late August. It’s now reportedly in talks for four times that. Nothing is signed yet: the round hasn’t closed, and the number could still move. The company’s [previous $250 million Series B](https://techfundingnews.com/noah-shinns-instinct-goes-from-100m-to-2-5b-in-weeks-23-year-olds-ai-assistant-just-raised-250m-series-b/) was co-led by Index Ventures and Benchmark, taking total funding to $350 million. ## The pattern of rapid ascent Cognition, the company behind AI coding agent Devin, [raised $2 billion at a $48 billion valuation in September](https://techfundingnews.com/devin-maker-cognition-raises-over-2b-at-48b-valuation-nearly-doubling-its-price-tag-in-four-months/), just four months after its [previous $26 billion valuation](https://techfundingnews.com/cognition-ai-eyes-40b-valuation-less-than-three-months-after-26b-mark/). Its annualised run-rate revenue grew from $492 million to nearly $900 million over the same stretch. [Clay raised $115 million at a $7.1 billion valuation](https://techfundingnews.com/clay-hits-7-1b-valuation-with-115m-raise-as-it-sees-ai-agents-running-future-of-sales/), up from $3.1 billion a year earlier. That’s the pattern, three companies deep now. ## The valuation versus economics question - [AI](https://techfundingnews.com/tag/ai/) - [funding](https://techfundingnews.com/tag/funding/) - [news](https://techfundingnews.com/tag/news/) - [top funding rounds](https://techfundingnews.com/tag/top-funding-rounds/) ##### Related Posts - [AI News](https://techfundingnews.com/category/ai/) - [Startups News](https://techfundingnews.com

Opinion | My AI assistant is deeply annoying — and that’s the least of it

Recent reports of rogue artificial intelligence agents and warnings of human extinction quickly turned our dalliance with AI into a doomsday nightmare.

Prophix Unveils Next-Gen AI Agents for Prophix One Autonomous Finance Platform

 For fintech professionals, this represents a significant shift toward "autonomous finance," where governed AI agents handle recurring workflows and live data integration to reduce manual reporting burdens and improve strategic oversight.

Migrating multi-model AI agents to Amazon Bedrock AgentCore runtime

# Migrating multi-model AI agents to Amazon Bedrock AgentCore runtime ## Migrate the agent to AgentCore runtime ### Test the deployed agent ```python import boto3, json client = boto3.client('bedrock-agentcore', region_name='us-west-2') payload = json.dumps({ "prompt": "What are the side effects of metformin?", "model_type": "llama" }) response = client.invoke_agent_runtime( agentRuntimeArn='<your-agent-runtime-arn>', contentType='application/json', accept='application/json', payload=payload.encode('utf-8') ) result = response['response'].read().decode('utf-8') print(result) ``` ## Clean up ```bash agentcore remove all ``` ```bash agentcore deploy ``` ## Conclusion In this post, we showed how to migrate a multi-model healthcare AI agent from self-managed Amazon ECS with AWS Fargate infrastructure to Amazon Bedrock AgentCore runtime. ## About the author ### Resources - [What's New](https://aws.amazon.com/new?sc_ichannel=ha&sc_icampaign=acq_awsblogsb&sc_icontent=machine-learning-resources)

The internet belongs to the AI agents now

# The internet belongs to the AI agents now [](https://www.businessinsider.com/author/alistair-barr) Sep 18, 2026, 11:58 AM ET I visited the CEO of a major startup in SF last week. His company recently put a new product online but hadn't done a big launch or any marketing. Within a week or two, it got at least a million signups. Barely any humans even knew about it. This was all AI agents just finding this new tool and using it en masse. Silicon Valley got busy [building the "rails"](https://www.businessinsider.com/agentic-internet-ai-rails-google-anthropic-openai-2025-9) for this agentic web so that these autonomous software programs could act on our behalf, executing workflows, doing online chores, and even paying for things, all without us lifting a finger. Meta recently launched a personal agent called Muse, which got busy scouring the web and calling customer service on behalf of millions of users. It [hit the top of the app charts](https://www.businessinsider.com/meta-muse-personal-ai-agent-top-app-store-charts-2026-9) late Thursday. [Instinct](https://www.businessinsider.com/instinct-is-the-ai-assistant-silicon-valley-cant-stop-talking-2026-9), an agent that lives in your text messages, is performing similar feats. In the workplace, [coding agents](https://www.businessinsider.com/ai-coding-agents-cursor-human-review-2026-6) are churning out millions of lines of code faster than any human software engineer. ## Muse saves Since Meta's Muse launched last week, early users have been [recounting stories of the agent](https://www.businessinsider.com/mark-zuckerberg-muse-ai-meta-baking-tips-mma-advice-2026-9) going out on the web to find and negotiate new subscriptions, insurance policies, and other cost savings. Joseph Devoy, founde

Collibra Survey Finds 76% of Organizations Hit Roadblocks Scaling AI Agents

NEW YORK and BRUSSELS, Sept. 18, 2026 — Collibra has announced the findings of a Harris Poll survey of more than 300 U.S. adults ages 21+ who are employed...

Iran and China Create First-of-Their-Kind Autonomous A.I. Influence Campaigns

The countries, along with Israeli firms, combined Chinese open-source A.I. models and A.I. agents for the campaigns, foreshadowing future online...

War in the Middle East & Rising Interest Rates Threaten the Funding for AI Build-Out

# War in the Middle East & Rising Interest Rates Threaten the Funding for AI Build-Out ## As Oil Revenues Slide, Middle Eastern Wealth Funds Stick With AI — for Now Just this month, [Saudi Arabia pledged $15 billion](https://startupfortune.com/saudi-arabia-pledges-over-15-billion-in-ai-deals-at-leap-2026-in-riyadh/) in domestic AI investment. ## Meta’s Muse Eats into Instinct’s Early Lead But even as Benchmark-backed Instinct [reportedly seeks a $10 billion valuation](http://theinformation.com/articles/ai-agent-startup-instinct-talks-10-billion-valuation?rc=fn2ric), Meta is nipping at its heels.

Inside Temporal's journey to a $12.55B valuation, and its bet on building a backbone for AI agents

The founders also see an opening in how AI systems connect to one another.