--- title: "Companies Cannot Track Their Own AI Agents | Metaintro" canonical: "https://www.metaintro.com/blog/companies-cannot-track-ai-agents-new-jobs" language: "en" author: "drashtigarach" published: "2026-06-17T13:43:15.000Z" modified: "2026-06-17T13:43:15.000Z" --- [Back to Blog](/blog) [AI](/blog/tag/ai)[Technology](/blog/tag/technology)[Jobs](/blog/tag/jobs)[Career](/blog/tag/career)[Skills](/blog/tag/skills) # Companies Cannot Track Their Own AI Agents, the New Jobs This Gap Creates Only 42% of IT teams know who owns each AI agent. See the new AI governance, AgentOps, and security jobs this control gap creates, and how to land one. [![Drashti Garach](https://cdn.metaintro.com/rs:fill:40:40/q:72/plain/images/5719d740-e510-42bc-8017-e040d145f35f_1766029465094.png)Drashti Garach @DrashtiGarach](/blog/author/drashtigarach) [June 17, 2026](/blog/archive/2026/06)12 min read ![Companies Cannot Track Their Own AI Agents, the New Jobs This Gap Creates](https://cdn.metaintro.com/rs:fill:1200:675/q:78/plain/images/companies-cannot-track-ai-agents-new-jobs-hero.png) [https://x.com/intent/tweet?text=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs](https://x.com/intent/tweet?text=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs&title=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs&title=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates)[mailto:?subject=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs](mailto:?subject=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs) Companies are racing to deploy AI agents, but most cannot say who owns them or what they can access. A [VentureBeat](https://venturebeat.com/) report on a survey of 1,500 IT professionals found that 85% of teams claim every AI agent is under control, yet only 42% actually know who owns each one. That gap is not just a security headache. It is creating real jobs in AI governance, AgentOps, and AI security, with strong salaries. At [Metaintro](https://www.metaintro.com), we track how shifts like this turn into hiring demand, so you can move toward the roles companies are scrambling to fill. ## What Did the Survey Actually Find? The headline numbers come from a study of 1,500 IT professionals, administered independently by Ravn Research and MSI Advanced Customer Insights and reported by [VentureBeat](https://venturebeat.com/). The standout finding is a confidence gap. While 85% of IT teams insist every AI agent is under control, only 42% can name who owns each agent. Ownership matters because an agent with no clear owner is an agent no one is patching, monitoring, or shutting down when it misbehaves. The same study found that among companies with AI policies in place, just 24% of employees said those policies are followed "very consistently" day to day. In other words, the rules exist on paper, but enforcement is thin. The findings were discussed at the RSA Conference, where security leaders, including the chief technology officer of [CrowdStrike](https://www.crowdstrike.com/), warned that autonomous agents are multiplying faster than the teams meant to supervise them. This is the same trust gap we cover in our look at [why almost every company runs AI agents but only a fraction trust them enough to ship](https://www.metaintro.com/blog/enterprise-ai-agents-trust-gap-2026-jobs). This is not an isolated data point. A separate survey from the [Cloud Security Alliance](https://cloudsecurityalliance.org/), commissioned by Token Security, found that 82% of enterprises have unknown AI agents running in their environments, and 65% experienced an AI agent related incident in the past 12 months. Of those incidents, 61% led to data exposure and 35% caused financial losses. Only 21% of organizations reported having a formal process for retiring or decommissioning agents once they are no longer needed. ## Why Is Agent Sprawl Happening So Fast? The root cause is speed. AI agents are cheap to spin up and easy to copy. A single employee can launch a half dozen agents to draft emails, query databases, or move data between systems, often without telling anyone in IT. Each agent typically holds credentials and permissions so it can act on its own. When an agent is forgotten, those permissions linger. The Cloud Security Alliance describes this as "retirement debt," where decommissioned or abandoned agents keep their access long after anyone remembers they exist. Compare this to how companies manage human employees. When a person leaves, HR and IT revoke their badge, email, and system access in a coordinated process. AI agents have no equivalent offboarding workflow at most companies yet. They are created in minutes and forgotten in days, but they keep operating. Multiply that across thousands of employees and you get the sprawl the survey describes: a fleet of digital workers no single team can fully see. The math gets worse the longer it goes unmanaged. An agent built to reconcile invoices may quietly retain read and write access to a finance system for a year after the project that created it ended. Another, set up to summarize support tickets, may still hold a live key to a customer database that no one has rotated. Because each agent looks small on its own, none of them trips an alarm, yet together they form a shadow workforce with real reach into sensitive systems. Security teams often discover the scale only during an audit or after an incident, when they finally inventory what is connected and find dozens of agents nobody can account for. For job seekers, this messy reality is good news. Every problem a company cannot solve internally becomes a role it needs to hire for. The agent governance gap is large, expensive, and growing, which is exactly the kind of pain that opens new job categories. ## Which New Jobs Is This Gap Creating? The control gap is spinning up several distinct roles. Understanding what each one does helps you target the right one for your background. - **AI governance lead or AI governance manager.** This person owns the policies, ownership registry, and approval workflows for AI agents. They decide who can deploy an agent, what it can access, and how it gets retired. Strong fit for people from compliance, risk, audit, or program management backgrounds. - **AgentOps or AI operations engineer.** A technical role focused on deploying, monitoring, and decommissioning agents at scale. Think of it as the platform team for AI workers. This suits people coming from DevOps, MLOps, or site reliability engineering. - **AI security engineer or analyst.** This role hunts for unknown agents, locks down their permissions, and responds when an agent leaks data or behaves unexpectedly. A natural next step for cybersecurity analysts and identity access specialists. - **AI risk and compliance specialist.** This person maps AI agent use against regulations and internal standards, and prepares the company for audits. A fit for legal, GRC, and privacy professionals. The common thread is oversight. Companies do not just need more people building agents. They need people who can see, name, secure, and shut them down. That is a different skill set, and it is in short supply. The same shift is reshaping technical careers, as we explain in our piece on how [software engineers are becoming AI agent managers](https://www.metaintro.com/blog/software-engineers-ai-agent-managers-how-to-stay-ahead). ## What Do These Roles Pay? The pay for oversight roles is strong, and rising, because the talent pool is thin. While "AI governance lead" is too new to have its own [Bureau of Labor Statistics](https://www.bls.gov/) wage code, the adjacent occupations that feed these jobs are well documented and give you a reliable salary floor. Information security analysts, the closest match for AI security roles, earned a median annual wage of $124,910 in May 2024, according to the [Bureau of Labor Statistics](https://www.bls.gov/). The top 10% earned more than $186,420, and the field is projected to grow 29% from 2024 to 2034, far faster than the average for all occupations. AI security work, which is more specialized and harder to staff, tends to land at the upper end of that range or above. For a sense of how cybersecurity demand is moving, see our coverage of [what recent security layoffs mean for cybersecurity jobs](https://www.metaintro.com/blog/sentinelone-cuts-8-percent-workforce-2026-cybersecurity-jobs). On the technical side, data scientists earned a median of $112,590 in May 2024, with the top 10% above $194,410 and projected job growth of 34% over the decade. Computer and information research scientists, a category that often includes advanced ML and AI roles, posted a median of $140,910, with top earners above $232,120. These figures matter because AgentOps and MLOps roles draw directly from this talent, and specialized AI experience pushes offers higher. It helps to read these numbers as a floor rather than a ceiling. The salary surveys describe established occupations, while agent oversight layers a scarce new specialty on top of them. When a role asks for security or data fundamentals plus hands on experience governing autonomous agents, employers are competing for a much smaller pool, and compensation reflects that scarcity. The takeaway for your career: oversight and security skills are not a side track. They sit among the better paid corners of the technology job market, and the agent governance gap is adding demand on top of an already tight field. ## How Do You Know These Jobs Are Actually Real? It is fair to be skeptical of a brand new job category. The strongest evidence is that the spending and the postings are already showing up, not just the survey talk. The same companies that admit they cannot track their agents are the ones writing checks to fix it, and budget is the clearest signal that hiring follows. Watch three concrete signals as you job hunt. First, look for net new titles rather than relabeled old ones, such as AI governance, AgentOps, or AI security analyst, which tells you a team is being built from scratch. Second, read the requirements section closely, because postings that pair a security or compliance base with explicit mentions of autonomous agents, model risk, or AI inventory are the real oversight roles. Third, note who is hiring, since vendors selling agent security tooling and the enterprises buying that tooling tend to staff up at the same time. Even where a dedicated title does not exist yet, the work is being absorbed into adjacent jobs. A security analyst may pick up agent monitoring, or a compliance manager may start owning the AI policy. If you can do that work today, you are already a candidate, and you are positioned to step into the dedicated role the moment a company formalizes it. ## What Does This Mean for Your Career? Here is how to position yourself for these roles, whatever your starting point. - **Learn a recognized AI governance framework.** Study the [NIST](https://www.nist.gov/) AI Risk Management Framework. It is free, widely referenced, and gives you the vocabulary hiring managers use for agent oversight. - **Build an identity and access mindset.** The core of agent control is knowing who owns what and what each agent can touch. Skills in identity access management, least privilege, and credential management transfer directly. - **Get hands on with agent tooling.** Experiment with agent platforms and document how you would track, monitor, and retire an agent. A simple "agent inventory" project on your resume signals you understand the real problem. - **Pair a soft skill with a technical one.** Governance roles reward people who can write a clear policy and read a log file. If you come from compliance, add technical literacy. If you come from engineering, learn risk and audit basics. - **Target companies feeling the pain.** Firms deep into AI adoption are the ones hiring oversight roles first. Watch for job titles like AI governance, AgentOps, AI security, and AI risk. Earning a credential can also help you stand out, as we detail in our guide to [how AI certifications boost salary and career growth](https://www.metaintro.com/blog/ai-certifications-boost-salary-and-career-growth-in-2026). Lacey Kaelani, CEO of [Metaintro](https://www.metaintro.com), told [Fortune](https://fortune.com/), "The candidates who win in this market are the ones who learn to manage and direct AI, not compete with it. Companies will always pay for people who can keep their technology accountable." That is the mindset these roles reward. ## How Should You Reframe AI Anxiety Into Opportunity? It is easy to read agent sprawl as another sign that AI is coming for jobs. The data points the other way. Every wave of automation creates a parallel demand for people to govern, secure, and correct it. The companies in these surveys are not short on agents. They are short on humans who can keep those agents in line. If you have spent months worried that AI will replace your role, consider flipping the script. The same technology that feels threatening is generating brand new job categories that did not exist two years ago. Oversight work is durable because it grows with adoption. The more agents a company runs, the more it needs people to watch them. At [Metaintro](https://www.metaintro.com), we see this pattern repeat across the labor market: disruption opens doors for those who position early. --- ## Related Articles - [HR Now Manages AI Agents Too, the New Skills That Make You Hireable in 2026](https://www.metaintro.com/blog/hr-manages-ai-agents-skills-hireable-2026) - [AI Agents Are Not Your New Coworkers, 2026 Research Maps the Failure Modes Managers Ignore](https://www.metaintro.com/blog/ai-agents-not-employees-hbr-research-2026) - [The 85/5 Enterprise AI Paradox, Why Almost Every Company Runs AI Agents but Only 5% Trust Them Enough to Ship](https://www.metaintro.com/blog/enterprise-ai-agents-trust-gap-2026-jobs) - [TCS Will Have as Many AI Agents as Employees Within Three Years, What IT Workers Should Do Now](https://www.metaintro.com/blog/tcs-ai-agents-match-workforce-2026) - [SentinelOne Cuts 8 Percent of Its Workforce in 2026 and What It Means for Cybersecurity Jobs](https://www.metaintro.com/blog/sentinelone-cuts-8-percent-workforce-2026-cybersecurity-jobs) - [514,000 Cybersecurity Jobs Open in 2026 as Security Becomes Tech's Most Resilient Career Path](https://www.metaintro.com/blog/cybersecurity-jobs-open-2026-tech-resilient-career) - [AI Certifications Boost Salary and Career Growth in 2026](https://www.metaintro.com/blog/ai-certifications-boost-salary-and-career-growth-in-2026) - [62% of Employers Cannot Find Workers With AI Skills and Training Is Getting Worse](https://www.metaintro.com/blog/ai-skills-gap-employers-hiring-2026) - [Your New AI Coworker Just Got Hired, 7 Skills You Need to Thrive Alongside AI Agents in 2026](https://www.metaintro.com/blog/ai-agents-workplace-skills-thrive-career-2026) --- ## People Also Asked ### Q: What is an AI agent in simple terms? A: An AI agent is a piece of software that can take actions on its own to complete a task, such as pulling data, sending messages, or updating records. Unlike a basic chatbot that only responds, an agent can string together steps and act across systems. That autonomy is why companies struggle to track them and why oversight roles are growing. ### Q: Do I need to be a programmer to work in AI governance? A: Not always. Pure AI security and AgentOps roles are technical and reward coding and systems skills. But AI governance, risk, and compliance roles lean heavily on policy, audit, and communication abilities. Many people enter from legal, compliance, project management, or operations backgrounds and add technical literacy over time. ### Q: How do I get my first AI oversight role with no direct experience? A: Start by translating skills you already have. If you work in security, compliance, or operations, you likely already do parts of this job. Learn the [NIST](https://www.nist.gov/) AI Risk Management Framework, build a small portfolio project around tracking or auditing AI agents, and apply to roles that list AI governance or AI security in the description. Future-proof your career by moving toward the roles this gap is creating, not away from the technology behind it. The companies struggling to track their AI agents need people who can govern, secure, and direct them, and those skills are learnable starting today. [Sign up with Metaintro](https://www.metaintro.com/signup) to get matched with AI governance, AgentOps, and AI security roles as companies race to fill them. ### Share this article [https://x.com/intent/tweet?text=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs](https://x.com/intent/tweet?text=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs&title=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs&title=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates)[mailto:?subject=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs](mailto:?subject=Companies%20Cannot%20Track%20Their%20Own%20AI%20Agents%2C%20the%20New%20Jobs%20This%20Gap%20Creates&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompanies-cannot-track-ai-agents-new-jobs) For job seekers ## Spend your time on the work, not the search. Free to start. No résumé needed. Real jobs, picked for you, by text. [Get Started Free](/signup)[Search matching jobs](/jobs/search) - MatchCompare live roles against your current evidence. - PositionTurn proof projects into role-specific applications. - ImproveUse market feedback to keep the skill plan current. [Return to navigation](#main-navigation)