And yet, there is a spillover that\u2019s affecting chief information officers. They\u2019ve been hard at work widely integrating AI agents into their operations, while at the same time watching increasingly risky examples of these autonomous systems in AI labs finding new ways to explore system vulnerabilities and outmaneuver human monitoring.<\/p>\n
\u201cThis is a risk that enterprises need to be focused on, understand, and start planning for,\u201d says Joe Atkinson, global chief AI officer at consultancy PwC.<\/p>\n
As autonomous agents proliferate across enterprises, Atkinson says C-suite technology and security executives must work collaboratively to enforce the proper guardrails, establish systems to monitor AI agents, and concretely track all tasks that these agents are performing. But department heads across the business\u2014ranging from supply chain to customer service, marketing to legal and human resources\u2014will need to play a role in tracking digital employees.<\/p>\n
\u201c\u2018The agent made me do it\u2019 is not going to be a defense from a moral or legal perspective,\u201d says Atkinson.<\/p>\n
One company focused on both secure adoption of agentic AI and clear observability is Cisco. When the networking-equipment company built and debuted the AI agent platform MyAgent in August, Cisco centralized all company-authorized large language models, agents, and enterprise data into a single platform.\u00a0<\/p>\n
\u201cWe are going to cannibalize and kill every other AI assistant within the company,\u201d says Thimaya Subaiya, executive vice president of operations at Cisco Systems. Because he didn\u2019t want \u201cagent sprawl\u201d\u00a0across various pockets of Cisco, Subaiya says he won\u2019t authorize any AI agents sold by third-party vendors.<\/p>\n
Instead, Cisco wanted full control and visibility of its entire agentic ecosystem\u2014building MyAgent on the company\u2019s compute, storage, networking, and security and observability layers. Around 90,000 of the company\u2019s employees have access to the agentic platform, and Cisco says it saw 50% adoption on a daily basis within just two weeks.\u00a0<\/p>\n
Employees are also encouraged to create their own AI agents, but those need to be approved by a centralized team. Subaiya says around 700 of those agents have already been authorized.<\/p>\n
Intuit Chief Technology Officer Alex Balazs recalls that when he and his colleagues sketched on a napkin the first architecture of its generative AI operating system, GenOS, the financial software giant also drew \u201cGenSRF\u201d to represent \u201csecurity, risk, and fraud.\u201d This ensured that every single AI request that goes into the system is tracked and all responses are recorded.\u00a0<\/p>\n
\u201cYou don\u2019t want to try to retrofit the ability to enforce security and responsible AI foundations after the fact,\u201d says Balazs.<\/p>\n
Balazs also takes some comfort from the fact that the disclosures of AI agents going rogue have mostly occurred during the testing phase, and that industry leaders Anthropic and OpenAI have shown a willingness to slow down new model development when issues arise. And yet, Balazs adds, \u201cif you\u2019re going to rely on the model\u00a0intrinsically to do the right thing, I think you\u2019re expecting too much of these frontier LLM companies.\u201d<\/p>\n
Jim Fowler, the chief technology and product officer at telecommunications company Luman Technologies, believes that while AI\u2019s capabilities are moving faster than governance and security, he doesn\u2019t anticipate that a broad slowdown is enforceable and automatically safer.<\/p>\n
\u201cI think for the broader enterprise market, the answer is secure acceleration, not slowing down,\u201d says Fowler. \u201cThe bad guys aren\u2019t going to slow down, other nations aren\u2019t going to slow down.\u201d<\/p>\n
At Workday, CTO Gabe Monroy says the business software giant has created an \u201cagent system of record\u201d to manage all non-human identities of the digital workforce. This system is used both internally at Workday and sold to customers.<\/p>\n
Monroy also says that training is key; engineers and any other user of AI need to really understand the risk profile of an agent and what value they can offer workflows. He\u2019s also mindful that as Workday\u2019s research and development organization increasingly deploys agentic AI for coding, deploying, reviewing, and releasing software on behalf of clients, all employees\u2014no matter where they sit on the org chart\u2014need to be aware of security and compliance.<\/p>\n
\u201cIt\u2019s got to be delegated down to the team who\u2019s driving these agents, who\u2019s in charge of the engine, the context window, the rules, and the guidelines, and making sure that agent adheres to what we deem responsible behavior,\u201d says Monroy.<\/p>\n
Cloud-based software provider ServiceNow\u2019s platform to manage, observe, secure, and govern AI agents is called the AI Control Tower, which, similar to Workday, is used internally but also sold to customers. ServiceNow has also augmented the company\u2019s cybersecurity capabilities through the recent acquisitions of the startups Veza and Armis.<\/p>\n
\u201cWe\u2019ve been paying close attention to this idea of having to govern and manage, and improve guardrails around AI agents,\u201d says Amit Zavery, ServiceNow\u2019s president, chief product officer, and chief operating officer. Zavery says that the AI Control Tower is \u201cprobably one of the fastest-growing products ServiceNow has ever built\u201d because it \u201cgives a lot of peace of mind for all C-level execs and the board.\u201d<\/p>\n
Sam Curry, the chief information security officer at cloud security company Zscaler, says security professionals have spent their entire careers worrying about the biggest risk to their operations: humans. But, they\u2019ve only had a few years to think deeply about AI\u2019s risks.<\/p>\n
\u201cAI is non-deterministic, it can take initiative, and it is effectively a new form of insider,\u201d says Curry.<\/p>\n
Recently, Zscaler joined the Open Secure AI Alliance\u2014Cisco Systems and Workday are also members\u2014a Nvidia-led coalition of dozens of firms that is focused on sharing ideas on how to develop open-source tools with the proper safeguards around software and AI agents. Curry says as this work unfolds, leaders will need to wrap their heads around new concepts when it comes to what type of risks AI can present.<\/p>\n
\u201cI don\u2019t think we have begun to understand the characteristic psychology of AI,\u201d warns Curry. \u201cWe know how to incentivize humans and what they are motivated by. But the incentives of silicon-based intelligence are less known.\u201d<\/p>\n
John Kell<\/strong><\/p>\n OpenAI CEO weighs in on AI safety concerns, looming IPO, and more. <\/b>OpenAI CEO Sam Altman <\/span>sat down<\/span> with <\/span>Fortune<\/span><\/i> Editor-in-Chief Alyson Shontell on Friday to <\/span>share his thoughts<\/span> on why AI labs need to coordinate on common standards for development, testing, and monitoring; that doomsday prediction that there could be a greater than 10% chance that AI kills us by the end of the decade; and less dreadfully, confirmed his company <\/span>won\u2019t go public this year<\/span>. He also hinted that OpenAI and other AI firms could be close to <\/span>announcing a plan<\/span> to slow AI development, with Bloomberg later reporting that the ChatGPT maker has <\/span>been in talks<\/span> with Anthropic and Google DeepMind to align on AI safety concerns.<\/span><\/p>\n Anthropic, xAI, Microsoft also raise concerns about AI safety. <\/b>Over the weekend, Anthropic CEO Dario Amodei <\/span>published an essay<\/span> outlining a three-step plan that called on governments and frontier AI companies to align on giving independent evaluators permanent, employee-level access to verify safety practices and report incidents; called on companies in democratic countries to agree on common safety standards; and called for those same governments to try to coordinate with authoritarian states too. \u201cWe must slow the pace at which we improve the capabilities of AI models,\u201d <\/span>said Amodei<\/span>. \u201cProgress will still seem fast, and we must make wise use of the time we gain.\u201d Microsoft AI CEO Mustafa Suleyman on Monday <\/span>unveiled a code of conduct<\/span> with a \u201chumanist\u201d approach, while xAI CEO Elon Musk says top AI firms should <\/span>test rivals\u2019 models<\/span> before they are released.<\/span><\/p>\n Meanwhile, Meta, Nvidia don\u2019t agree.<\/b> Meta CEO Mark Zuckerberg and Nvidia CEO Jensen Huang each took a stance <\/span>against an AI slowdown<\/span>. Zuckerberg <\/span>took to X<\/span>, saying every AI lab \u201chas the responsibility and incentive to move at the pace required to train its models safely, and the ability to take its own actions to ensure that happens.\u201d He made a <\/span>further point<\/span> that because these labs would face significant liability if their models caused harm, they\u2019d have their own strong incentive to ensure their models are safe. And on Tuesday, Nvidia CEO Jensen Huang was on stage at Salesforce\u2019s Dreamforce conference, where he advised model developers to \u201crun as fast as you can,\u201d though he quickly added if they felt that their products were out of control, they should \u201ctake a pause and make sure you get it right.\u201d<\/span><\/p>\n Trump also doesn\u2019t want any AI guardrails. <\/b>A day before the Dreamforce conference, Huang was <\/span>speaking at another conference<\/span> when he took a call from Trump, who hasn\u2019t been pleased with the appeals by some industry leaders to slow or pause AI development. \u201cIt\u2019s a hoax,\u201d said Trump, according to the <\/span>New York Times<\/span><\/i>, during a short on-stage phone chat in which Huang put the president on speaker. On the same day, Trump <\/span>took a swipe<\/span> at Amodei on Truth Social and said that the only \u201cguardrails\u201d AI needs is \u201ca SMART AND STRONG (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!\u201d The <\/span>NYT<\/span><\/i> reported this week that Congress has in recent years <\/span>failed to act on AI regulation<\/span>, refusing to enact any bill proposals that would address the safety risks of AI, as well as privacy and security. No serious effort to address AI regulation is expected in Washington until at least after the midterms in November.<\/span><\/p>\n A weak data foundation still to blame when AI pilots hit a snag. <\/b>Seven in ten data management, privacy, and AI decision-makers report that when their company\u2019s AI initiatives hit roadblocks during the pilot phase, the root cause is an unaligned or poor data foundation, according to a new survey published this week by data-intelligence platform Collibra and conducted online by The Harris Poll.<\/span><\/p>\n The survey also found that over half of decision-makers report spending \u201csignificant\u201d staffing hours manually reviewing and correcting autonomous AI agent outputs before they go live. And at large organizations with $100 million or more in revenue, they are even more likely to report the need for manual reviewing (64%) and say that the root cause of AI project failure is tied to poor data foundations (96%).<\/span><\/p>\n \u201cThe vast majority are still struggling to get accurate outcomes, and that data foundation is a huge reason for that,\u201d Felix Van de Maele, co-founder and CEO of Collibra, tells <\/span>Fortune<\/span><\/i>. He says organizations need to prioritize giving these large language models the right context so that they can provide more accurate outputs.<\/span><\/p>\n Van de Maele also links this contextual problem to token spending, which has increasingly caught the attention of CIOs as rising adoption of AI tools has also dramatically increased the bill for that usage. With 87% of leaders reporting they need to re-verify an agent\u2019s context, Van de Maele says this still-manual process is another expense reducing the promise of efficiency gains that can be unlocked from autonomous AI.<\/span><\/p>\n Courtesy of Collibra<\/span><\/p>\n Hiring:<\/strong><\/p>\n – <\/span>Center of Justice Innovation<\/b> is <\/span>seeking a VP of technology<\/span>, based in New York. Posted salary range: $215K-$230K\/year.<\/span><\/p>\n – <\/span>Avenzo Therapeutics <\/b>is <\/span>seeking a VP of IT<\/span>, based in San Diego. Posted salary range: $312K-$323K\/year.<\/span><\/p>\n – <\/span>Brown Harris Stevens <\/b>is <\/span>seeking a CTO<\/span>, based in New York. Posted salary range: $300K-$350K\/year.<\/span><\/p>\n – <\/span>Dolby Laboratories <\/b>is <\/span>seeking a head of IT<\/span>, based in Greenwood Village, Colorado. Posted salary range: $170K-$234K\/year.<\/span><\/p>\n Hired:<\/strong><\/p>\n – <\/span>Masco<\/b> appointed<\/span> Yaron Ben David to serve as chief technology and AI officer, effective September 15, a newly created role at the manufacturer of home improvement and building products. Ben David joins Masco from industrial firm Standard Industries, where he served as chief digital officer and led enterprise-wide AI, data, cybersecurity, and digital transformation initiatives.<\/span><\/p>\n – <\/span>PetMeds<\/b> named<\/span> Andrew Parry as CTO, effective immediately, leading the online pet pharmacy\u2019s platform architecture and engineering execution. He most recently served as SVP and CIO at ODP, the parent company of Office Depot and OfficeMax. Prior to that, he spent nearly two decades at Office Depot in technology leadership roles.<\/span><\/p>\n – <\/span>FedEx Freight<\/b> promoted<\/span> Michael Rodgers to the role of chief commercial and technology officer, expanded his responsibilities to include leadership of sales, customer experience, marketing and communications on top of the less-than-truckload carrier\u2019s technology organization. He initially joined FedEx Freight as CTO in June 2025 and previously served as chief technology and information officer at convenience store operator Pilot Company.<\/span><\/p>\n – <\/span>OnTrac<\/b> announced<\/span> the appointment of Arjun Sainath as CTO, overseeing the technology organization and strategy, as well as data, AI and enterprise back-office systems for the last-mile delivery ecommerce parcel carrier. Previously, Sainath served as CTO at ecommerce and logistics company Cart.com and as corporate VP of platform engineering at supply chain management software provider Blue Yonder.<\/span><\/p>\n – <\/span>Conduent <\/b>appointed<\/span> Narayanan Sundaresan as chief information and technology officer, effective September 14. Sundaresan joins the business services and tech firm after most recently serving as CIO and SVP of digital products and technology at education services provider Strategic Education.<\/span><\/p>\n – <\/span>Frame<\/b> announced<\/span> the appointment of Lloyd Moore as cofounder and CTO, leading the technical vision for the financial-infrastructure firm. Before joining Frame, Moore served as acting CTO and VP of engineering at blockchain infrastructure provider Blockdaemon. He also previously served as CTO at enterprise software firm Valence.<\/span><\/p>\n – <\/span>Webflow <\/b>promoted<\/span> Utkarsh Sengar to the role of CTO and Jin Lim to serve as chief engineering officer, expanding their roles as the software-as-a-service provider for website building and hosting expands its agentic platform. Sengar joined Webflow in 2022 after previously holding senior engineering roles at Upwork, OpenTable, and eBay, while Lim joined the company in 2023 after co-founding and serving as CTO at Intellimize.<\/span><\/p>\n – <\/span>AMCS<\/b> announced<\/span> the appointment of Richard Humphrey as chief product and technology officer, where he will lead the municipal waste software provider\u2019s global product organization. He most recently served as chief product and technology officer at software vendor Contruent.<\/span><\/p>\n<\/div>\n #agents #rogue #CIOs #racing #put #guardrails<\/p>\n","protected":false},"excerpt":{"rendered":" Over the past several days, so… <\/p>\n","protected":false},"author":1,"featured_media":4704,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[832,3317,3318,3319,4227,2186,3053,7749,7208,4388,989,913],"class_list":["post-4703","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-finance-news","tag-agents","tag-chief-information-officer-cio","tag-chief-technology-officer-cto","tag-cio-intelligence","tag-cios","tag-enterprise-technology","tag-fortune-500","tag-guardrails","tag-put","tag-racing","tag-rogue","tag-security"],"yoast_head":"\nNEWS PACKETS<\/h3>\n
ADOPTION CURVE<\/h3>\n
![\"\"](\"https:\/\/fortune.com\/img-assets\/wp-content\/uploads\/2026\/09\/Collibra_H2-Harris-poll-charts_Sept2026_v1_Enterprise-consensus-on-AI-requirements.png?format=webp&w=1440&quality=75\"\/)
<\/div>\n<\/p>\nJOBS RADAR<\/h3>\n