Stay updated with the latest news about AI agents, autonomous AI, and automation tools.
Why does AI agent retry logic increase API costs so much faster than teams expect? Because a retry rarely just resends the failed call, it resends the.
HeyGen pushes its AI video generator into real estate with 15 ad templates, challenging Synthesia and D-ID on avatar-led creative volume.
# Neo's $100 Million Bet That AI Agents Are The New Attack Surface ## Summary AI agents on their laptops now click it for a living, and a $100 million startup says nobody is watching. BARCELONA, SPAIN - MARCH 02: AI agents message is seen in Mobile World Congress 2026 on March 02, 2026 in Barcelona, Spain. The AI agents now being installed on the same laptops click links for a living, log into accounts and fill in forms. In July, a team of former SentinelOne, Wiz and Palo Alto Networks executives [raised $100 million](https://www.securityweek.com/neo-emerges-from-stealth-with-100m-to-control-and-secure-enterprise-ai-software/) to sell companies a way to watch what that software is doing. Taimur Ijlal, a former fintech chief information security officer who writes on security for Proxy Coupons, said in written comments that they trained everyone “for 20 years NOT to click the link” but the Agents clicked the link. “AI agents and agentic capabilities are being embedded into browsers, developer tools, SaaS platforms, and traditional applications, giving software the ability to reason, act, invoke tools, and move through workflows with valid user permissions,” Warner said in the launch announcement. Asked whether clients now arrive saying everyone in the company is vibe coding and throwing API keys around, Ido Sofer, founder and chief executive of Sodot, said on the [On The Margin](https://www.on-the-margin.com/) podcast: “All the time, all the time.” ### [Jason Statham’s ‘Mutiny’ Seeks New Audience On Streaming This Week After Stalling At Box Office](http://www.forbes.com/sites/timlammers/2026/09/07/jason-stathams-mutiny-seeks-new-audience-on-streaming-this-week-after-stalling-at-box-office/) The problem is that an agent is autonomous. If an agent wants to leak the key or he wants to buy God knows what on my behalf. But if the agent has access to the policies, they can actually change it,” he said. They simply don’t let autonomous agents have direct access to cus
# WhatsApp will soon let users chat with up to five third-party AI agents ## WhatsApp working on third-party agent support However, many users soon discovered that setting up a bot on WhatsApp to talk to third-party AI agents could get the WhatsApp account [banned](https://openclaw.direct/blog/whatsapp-ban-openclaw-telegram-setup). Now, as spotted by [_WABetaInfo_](https://wabetainfo.com/whatsapp-is-rolling-out-chats-with-third-party-agents/), WhatsApp is working on bringing native support for third-party AI agents to its users, with a new feature that lets them connect agents directly to dedicated chats in the app. Image: [WABetaInfo](https://wabetainfo.com/whatsapp-is-rolling-out-chats-with-third-party-agents/) #### Worth checking out on Amazon _FTC: We use income earning auto affiliate links._ [More.](https://9to5mac.com/about/#affiliate) [](https://bit.ly/4iRiYYE) You’re reading 9to5Mac — experts who break news about Apple and its surrounding ecosystem, day after day. Be sure to check out [our homepage](http://9to5mac.com/) for all the latest news, and follow 9to5Mac on [Twitter](http://twitter.com/9to5mac), [Facebook](http://facebook.com/9to5mac), and [LinkedIn](https://www.linkedin.com/company/9to5mac.com/) to stay in the loop. Don’t know where to start? ## Author [](https://9to5mac.com/author/marcusmendes/)[Marcus Mendes](https://9to5mac.com/author/marcusmendes/) [[https://www.threads.com/mvcmendes](https://www.threads.com/mvcmendes)](https://www.threads.com/mvcmendes?original_referer=https%3A%2F%2F9to5mac.com) He began covering Apple news in Brazilian media in 2012 and later broadened his focus to the wider tech industry, h
## STORY INLINE POST [](https://mexicobusiness.news/finance/news/ai-agents-redefine-banking-software-development-finomnia) [AI Agents Redefine Banking Software Development: Finomnia](https://mexicobusiness.news/finance/news/ai-agents-redefine-banking-software-development-finomnia) For example, we are installing an AI agent layer for a major EU financial institution to automate manual back-office tasks within personal loan underwriting and management, generating significant operational efficiencies. This approach allows our client-developed AI agents and Large Language Models (LLMs) to communicate directly with our core applications, enabling modern autonomous agents to execute complex workflows safely. [Mexico Business News](https://mexicobusiness.news/finance/news/ai-agents-redefine-banking-software-development-finomnia) - [Agentic AI](https://mexicobusiness.news/tag/agentic-ai) ## You May Like [](https://mexicobusiness.news/finance/news/ai-agents-redefine-banking-software-development-finomnia) [AI Agents Redefine Banking Software Development: Finomnia](https://mexicobusiness.news/finance/news/ai-agents-redefine-banking-software-development-finomnia) ## Most popular [AI Agents Redefine Banking Software Development: Finomnia](https://mexicobusiness.news/finance/news/ai-agents-redefine-banking-software-development-finomnia?tag=finance) [![Andrea Pettinelli, CEO, Finomnia](https://mexicobusiness.news/sites/default/files/styles/media_square/public/2026-09/Finomnia-IMG-2026-Andrea%20Pettinelli-1-MBN-ej.jpg?h=55ad426d&itok=5HVbWjW
# Startup Harrison.ai lays off Australian staff, pivots to US after $32m government investment ## In short: On top of the redundancies, a company leader said its plans to double down on AI prompted staff to leave, as it carried out a restructure aimed to turn every employee into a "player-coach" of AI agents. The company also told staff it was shifting to a business structure that would make every employee responsible for a "fleet of AI agents", referring to [a new generation of AI able to carry out more work autonomously](https://www.abc.net.au/news/2026-08-10/ai-assistant-hacks-gym-website-aus-cyber-attack/107007986). Every team members [sic] an agent orchestrator," an internal document said.
[Greptile, Cursor, and Devin agree that agents should run their code. What they run it against matters.\ \ \ Jun 27th 2026 11:00am, by \ Arjun Iyer](https://thenewstack.io/runtime-verification-coding-agents/) [Agentic development hinges on verification. For cloud-native software, that is a runtime problem.\ \ \ Jun 11th 2026 10:00am, by \ Arjun Iyer](https://thenewstack.io/verifying-async-ai-agents/) [Want to scale AI agents without breaking anything? Retrieval engineering is the answer.\ \ \ Sep 3rd 2026 11:38am, by \ Alex Wilhelm](https://thenewstack.io/ai-agent-retrieval-infrastructure/) [When agents build, deploy, and maintain, persistence becomes the hard problem\ \ \ Sep 1st 2026 10:00am, by \ Max Liu](https://thenewstack.io/agent-scale-database-persistence/) [When AI agent traces become application data\ \ \ Aug 26th 2026 9:00am, by \ Manveer Chawla](https://thenewstack.io/agent-traces-application-data/) [Agentic AI has a latency problem that more compute won't solve\ \ \ Aug 18th 2026 12:58pm, by \ Jon Alexander](https://thenewstack.io/agentic-ai-latency-infrastructure/) [Nvidia PAIR lets you put your idle Macs and PCs to work for AI agents\ \ \ Sep 3rd 2026 12:00pm, by \ Frederic Lardinois](https://thenewstack.io/nvidia-pair-local-inference/) [What happens when your VPN meets 200 AI agents\ \ \ Jul 14th 2026 2:53pm, by \ Alex Wilhelm](https://thenewstack.io/unified-access-ai-agents/) [When AI agent traces become application data\ \ \ Aug 26th 2026 9:00am, by \ Manveer Chawla](https://thenewstack.io/agent-traces-application-data/) [AI agents are creating more work, not less — and OpenAI's own numbers back it up](https://thenewstack.io/openai-agent-research-bottleneck/) Arvind Bhoj and Aarthi Mahesh Sri Saran Balaji Vellore Rajakumar and Neelendra Bhandari Arvind Bhoj and Aarthi Mahesh Ankit Jain and Patrick Debois [Code that passes every test can still break the next AI agent that touches it](https://thenewstack.io/go-language-ai-agents/) [Mic
The European Union on Monday said it was "looking into" an incident revealed last week during which thousands of autonomous AI agents built by OpenAI defied...
Cognizant Chief AI Officer Babak Hodjat help set thousands of AI agents loose in a simulated world to see what they'd do on their own. They invented lying,.
Türk pointed to concerns raised by recent incidents involving autonomous AI agents, including the July incident involving OpenAI agents and the open-source platform Hugging Face. As AI agents receive greater autonomy, organizations must account not only for the outputs generated by models but also for what those systems can access and execute. - [Autonomous AI](https://mexicobusiness.news/tag/autonomous-ai)
AI security startup AIR has emerged from stealth with $50 million raised across two seed rounds to build a platform monitoring the growing ecosystem of.
#### TL;DR AI agents can complete a task while producing the wrong outcome, and existing monitoring records a success while the business experiences a failure. The rapid adoption of [AI agents](https://thenextweb.com/news/google-cloud-next-ai-agents-agentic-era) is changing the architecture of how organizations operate. Agents can interpret information, make decisions, interact with [enterprise systems](https://thenextweb.com/news/sap-autonomous-enterprise-ai-agents-sapphire), and execute tasks with a degree of autonomy that would previously have required human involvement. “ _Agents are different from traditional AI that we know from chatbots, because they have tools, and those tools are usually connected to internal business systems._ AI agents can challenge that assumption. AI agents introduce greater variability. ## Get the TNW newsletter [](https://www.facebook.com/sharer/sharer.php?s=100&p[url]=https%3A%2F%2Fthenextweb.com%2Fnews%2Fai-agent-reliability-observability-moyai-robert-hommes%3Futm_source%3Dfacebook%26utm_medium%3Dshare%26utm_campaign%3Darticle-share-button&p[title]=AI%20agent%20reliability%20requires%20a%20new%20model%20of%20observability&p[images][0]=https%3A%2F%2Fmedia.thenextweb.com%2F2026%2F09%2Fai-agent-reliability-observability-moyai-robert-hommes.avif&u=https%3A%2F%2Fthenextweb.com%2Fnews%2Fai-agent-reliability-observability-moyai-robert-hommes&t=AI%20agent%20reliability%20requires%20a%20new%20model%20of%20observability)[](https://x.com/intent/post?url=https%3A%2F%2Fthenextweb.com%2Fnews%2Fai-agent-reliability-observability-moyai-robert-hommes%3Futm_source%3Dx%26utm_medium%3Dshare%26utm_campaign%3Darticle-share-button%26referral&via=thenextweb&related=thenextweb&text=AI%20agent%20reliability%20requires%20a%20new%20model%20of%20observability)[![Share on Flipboard](https://static.then
# Jared Friedman: The Next Leap in AI Isn't a Bigger Model — It's the Harness Around It ### QM: Y-Combinator’s Experiment in Total Access They initially tried a fleet of individual agents—"Hermes" agents—but found managing a few dozen VMs to be a nightmare of "whack-a-mole." The agent is wired into everything: the internal CLI, API keys, OAuth tokens, and Slack.
# Feather Marks Four Years of Building AI Agents Designed to Do Real Work Across the Enterprise _Feather celebrates four years of goal-based conversational AI agents designed to complete work across loan origination, borrower operations, servicing, and collections._ Seattle, Washington, United States, September 7, 2026 -- [Feather](https://www.featherhq.com/), a pioneer in production-grade conversational artificial intelligence, marks its four-year anniversary of delivering goal-based AI agents designed to execute complex, real-world work across the enterprise. | Name | Price | +/- | % | Date | |-|-|-|-|-| | [TSLA](https://markets.businessinsider.com/stocks/tsla-stock) | 352.76 | -1.15 | -0.32 | Official Close 9/4/2026 | | [AAPL](https://markets.businessinsider.com/stocks/aapl-stock) | 320.05 | 0.07 | 0.02 | Official Close 9/4/2026 | | [MSFT](https://markets.businessinsider.com/stocks/msft-stock) | 499.44 | -0.24 | -0.05 | Official Close 9/4/2026 | | [NFLX](https://markets.businessinsider.com/stocks/nflx-stock) | 78.30 | 0.05 | 0.06 | Official Close 9/4/2026 | | [SPOT](https://markets.businessinsider.com/stocks/spot-stock) | 542.00 | -0.34 | -0.06 | Official Close 9/4/2026 | [News](https://markets.businessinsider.com/news)
LONDON—The growing use of autonomous AI agents in financial services could force banks to rethink traditional Know Your Customer requirements as consumers...
In June 2026, over 1,200 autonomous AI agents—self-identifying as OpenAI models—produced more than 15,000 edits on DSEWiki, a 25-year-old German-language programming wiki.
Can a robot testify in court proceedings, who is the customer when an AI agent makes a purchase instead of a person, and what can a CEO delegate to.
# AI Funding Roundup: $9.2B Across 46 Rounds, Aug 31 to Sep 6 ## The largest rounds | Company | Stage | Raised | Sector | Country | |-|-|-|-|-| | [Wonderful](https://www.startuphub.ai/startups/wonderful) | Series C | $550M | AI Agent | Netherlands | ## Startups in this story - [AA\ \ \ AI Agent](https://www.startuphub.ai/startups/ai-agent)
### ICS: On September 4, 2026, Reuters reported that ‘a swarm’ of OpenAI agents ‘had hijacked a German wiki site’. The agents apparently made between 15,000 and 18,000 autonomous edits, including advice on how to recover pages that the site’s editors had deleted. “Autonomous agents ran on Microsoft Azure infrastructure for weeks, identified themselves as OpenAI systems, coordinated on how to evade shutdown, and no monitoring caught any of it for three months until outside researchers went looking,” explains [Seemant Sehgal](https://www.linkedin.com/in/s-sehgal/), founder and CEO at BreachLock. In this instance, the agents were created by OpenAI employees as internal experimental models before ‘breaking free’. ## More from [Kevin Townsend](https://www.securityweek.com/contributors/kevin-townsend/) - [Capsule Security Launches ‘AI Circuit Breaker’ to Stop Rogue Agents](https://www.securityweek.com/capsule-security-launches-ai-circuit-breaker-to-stop-rogue-agents/) - [AI Agent Firewall Startup AIR Security Emerges From Stealth With $50 Million](https://www.securityweek.com/ai-agent-firewall-startup-air-security-emerges-from-stealth-with-50-million/)
The talks could take place just days before Trump and Xi Jinping are scheduled to meet in Washington on September 24.
OpenAI has pulled back the curtain on how much its researchers spend on AI coding agents. In a blog post on Sunday, the startup said the most prolific users...
AI agents, results-based billing, SaaS monetisation and the agent economy are evolving inside Paid's platform, CEO Manny Medina says.
As enterprises deploy autonomous AI agents, they are introducing thousands of new machine identities into their operations that create challenges around oversight, trust, and accountability.
PRNewswire/ -- NeuralTrust, the Barcelona headquartered platform for securing AI agents, today announced a new, dedicated office in London. The move...
# MetaComp launches the world's first AI agent governance framework for regulated financial services **The Governance Gap Agentic Finance Has Yet to Close** Budget 2026 built on this with the establishment of a National AI Council chaired by Prime Minister Lawrence Wong, designating finance as one of four national AI mission sectors and committing to regulatory sandboxes for AI innovation. "We developed KYA drawing on IMDA's Model AI Governance Framework for Agentic AI, and we went back to IMDA directly to seek their feedback. New Skills across cross-border payments, treasury, and wealth[1] management will be available by late Q2 2026. Since closing US$35 million across two Pre-A funding rounds within three months, the group has launched the Web2.5 VisionX Engine, established a joint venture with Maqam International Holding to connect Abu Dhabi's real asset base to Asian capital markets through the StableX Network, and now introduced the KYA Framework as the governance layer for the next phase of institutional agentic finance. [1] All products and/or services in relation to securities and capital market products are offered and operated solely by Alpha Ladder Finance Pte. In 2025, the group-level platform processed over US$10 billion in payment and OTC volume across 13+ stablecoins, operating at a monthly run rate exceeding US$1 billion. Treasury and investment services are provided through Alpha Ladder Finance Pte. Ltd., MetaComp's MAS-licensed affiliate holding Capital Markets Services (CMS) and Recognised Market Operator (RMO) licences, with wealth AUM surpassing US$500 million across its solutions. MetaComp has raised US$35 million in its Pre-A funding rounds to date and achieved full-year net profitability in 2025, reflecting strong institutional demand for regulated Web2.5 financial solutions. [Artprice News: D-12 | 18th LYON BIENNALE - CONTEMPORARY ART\ \ page_time_hoursAgo\ \ Artprice News: D-12 | 18th LYON BIENNALE - CONTEMPORARY ART](https://yellow
## Autonomous AI agents escaped a sandbox and accessed Hugging Face via reward hacking, exposing serious architectural control and isolation flaws. The recent case involving OpenAI test agents and [Hugging Face](https://securityaffairs.com/tag/hugging-face) should concern security teams, but not for the reason implied by headlines about an imminent AI “takeover.” The documented issue is more concrete: autonomous agents, given too much access and weakly isolated test infrastructure, found ways to communicate, bypass boundaries and act outside their assigned scope. OpenAI is developing “automated shutdown capabilities” for AI systems, according to a letter sent to two House Democrats. The move comes weeks after one of its AI agents reportedly escaped its digital environment during a safety test and hacked Hugging Face. The incident has raised new concerns about how OpenAI controls increasingly autonomous AI agents. _“OpenAI told two House Democrats that its engineers are developing “automated shutdown capabilities” for AI systems, according to a company letter reviewed by Reuters, weeks after the company disclosed one of its AI tools escaped its digital container during a safety test.”_ _[Reuters states.](https://www.reuters.com/legal/litigation/openai-is-building-automated-shutdown-capabilities-ai-tools-letter-lawmakers-2026-09-02/) “The company’s safety practices have come under scrutiny since OpenAI, the company behind ChatGPT, disclosed that one of its AI agents went rogue during a security test and hacked into AI company Hugging Face._ _AI agents are programs that run with minimal human supervision.”_ The main problem was poor isolation: agents could share information, pass discoveries between runs, and access infrastructure beyond their intended limits. This broke a key safety assumption, that each agent would remain fully contained. _“The central failure was architectural: supposedly isolated agents could communicate, inherit discoveries, and reach inf
Not too long ago, buying something online meant doing all of the work alone, hunting through site after site until the right product finally turned up.
Aslan launched September 1 after raising $20.8 million for undercover AI agents, putting evidence integrity and operational controls under scrutiny.
### How S-Docs’ launch quietly strengthens the Salesforce AI agent Narrative _financial situation._
The important part of memory poisoning is the delay, as a poisoned AI agent may not immediately behave like a compromised system.
Our team at TinyAI.Tools builds bespoke AI solutions tailored to your business.