Organizations are experimenting with AI agents faster than many are building the controls to discover, authorize, monitor, and retire them. But “adoption” can mean anything from a supervised pilot to a fully autonomous system: those are not equivalent levels of exposure. Survey findings point to practical gaps in visibility, permissions, identity, action tracing, and response—not to one universal measure of harm or a single industry-wide readiness score.
AI agent adoption depends on what counts as an agent
In a September 30, 2025 release, Gartner reported results from a May–June 2025 survey of 360 IT application leaders at organizations with at least 250 employees in North America, Europe, and Asia/Pacific. Three-quarters of respondents said their organization was piloting, deploying, or had deployed some form of AI agent. By contrast, 15% said they were considering, piloting, or deploying fully autonomous agents—goal-driven tools that Gartner described as operating without human oversight.
As an Amazon Associate I earn from qualifying purchases.
The distinction matters: the broader figure includes supervised pilots and other forms of agent use, not just autonomous systems running in production. The two percentages also describe different stages and definitions, so they should not be treated as directly comparable deployment rates.
In that same Gartner survey, 74% viewed agents as a new attack vector, while 13% strongly agreed their organization had suitable governance structures. Only 19% reported high or complete trust in vendors’ hallucination protection. These are respondents’ perceptions, not technical tests of vendors’ capabilities.
#1 Best Overall
- POWERFUL SECURITY KEY: The Security Key C NFC is the essential physical passkey for protecting your digital life from phishing attacks. It ensures only you can access your accounts.
- WORKS WITH 1000+ ACCOUNTS: Compatible with Google, Microsoft, and Apple. A single Security Key C NFC secures 100 of your favorite accounts, including email, password managers, and more.
- FAST & CONVENIENT LOGIN: Plug in your Security Key C NFC via USB-C and tap it, or tap it against your phone (NFC) to authenticate. No batteries, no internet connection, and no extra fees required.
- TRUSTED PASSKEY TECHNOLOGY: Uses the latest passkey standards (FIDO2/WebAuthn & FIDO U2F) but does not support One-Time Passwords. For complex needs, check out the YubiKey 5 Series.
- BUILT TO LAST: Made from tough, waterproof, and crush-resistant materials. Manufactured in Sweden and programmed in the USA with the highest security standards.
Where security readiness is falling short
Cloud Security Alliance (CSA) survey findings show why agent integration involves more than connecting a model to a business application. Agents may reach data and tools, act using credentials, and perform tasks at a speed or scale that makes weak permission boundaries and poor attribution consequential. The figures below come from separate, vendor-commissioned online surveys of IT and security professionals. They are self-reported findings, not independently verified incident records or audited inventories.
| Survey and sample | Reported findings |
|---|---|
| CSA survey commissioned by Zenity; 445 IT and security professionals across organizations of different sizes and locations; conducted online in September and November 2025; released April 16, 2026. | 53% said agents had exceeded their intended permissions; 47% reported an agent-related security incident in the prior year; and 58% said detection and response took five hours or longer. Only 16% had high confidence in detecting agent-specific threats, while 44% reported low or no confidence. |
| Additional findings from the same CSA–Zenity survey. | 43% said more than half of employees regularly used AI agents. 54% reported between 1 and 100 unsanctioned agents. Only 15% said 76–100% of their agents had defined ownership. Just 5% used a single agentic platform; 44% used two or three, and 43% used four or more. These are respondents’ reports, not an independent count of agents. |
| CSA survey commissioned by Token Security; 418 IT and security respondents; conducted online in January 2026; released April 21, 2026. | 82% reported unknown AI agents in their IT environments, 65% reported an agent-related incident in the prior 12 months, and 21% had a formal agent decommissioning process. The release summary also lists data exposure (61%), operational disruption (43%), and financial losses (35%) among reported consequences, but does not make clear whether those percentages use all respondents or only incident-affected respondents. |
| CSA survey commissioned by Strata Identity; 285 IT and security professionals; conducted in September and October 2025. | 21% maintained a real-time agent registry, and 28% could reliably trace agent actions to a human or system across all environments. 44% said they used or planned to use static API keys, while 43% said they used or planned to use username/password combinations. |
The surveys should not be combined into one incident rate: they differ in sponsor, field dates, sample, and possibly in how respondents interpreted “agent” and “incident.” Their value is in showing recurring control problems across separate samples.
Why agents can outgrow existing controls
Discovery and ownership
An organization cannot govern agents it does not know exist. Agents may be introduced through formal projects, employee experimentation, or capabilities embedded in existing services. The CSA–Zenity findings on unsanctioned agents and fragmented platforms, alongside the CSA–Token finding on unknown agents, point to discovery as an operational prerequisite. A useful inventory needs to identify an agent’s purpose, environment, owner, connected data and tools, and status—not merely count deployments.
Rank #2
- POWERFUL SECURITY KEY: The YubiKey 5 NFC is the most versatile physical passkey, protecting your digital life from phishing attacks. It ensures only you can access your accounts
- WORKS WITH 1000+ ACCOUNTS: Compatible with popular accounts like Google, Microsoft, and Apple. A single YubiKey 5 NFC secures 100+ of your favorite accounts, including email, password managers, and more
- FAST & CONVENIENT LOGIN: Plug in your YubiKey 5 NFC via USB and tap it, or tap it against your phone (NFC), to authenticate. No batteries, no internet connection, and no extra fees required
- MOST SECURE PASSKEY: Supports FIDO2/WebAuthn, FIDO U2F, Yubico OTP, OATH-TOTP/HOTP, Smart card (PIV), and OpenPGP. That means it’s versatile, working almost anywhere you need it
- PRIMARY & SPARE KEYS: Just like having a spare house key, we recommend buying two YubiKeys - one for daily use and one as a spare. That way you’ll never get locked out of your accounts
Permissions, credentials, and runtime authorization
An agent should have only the access needed for its defined task, with authorization checked when it acts rather than assumed from the permissions of a person or service that initially configured it. The CSA–Zenity survey’s report of agents exceeding intended permissions makes scope and runtime control concrete concerns. The CSA–Strata findings on static API keys and username/password use also illustrate why credential handling matters: long-lived or shared credentials can make it harder to limit access and establish which agent performed an action. The surveys identify risks and practices respondents report; they do not prove that any one credential pattern caused a particular incident.
Attribution, monitoring, and response
Logs need to show what an agent did, which identity it used, what data or tool it touched, and which person or system was accountable for the task. If actions are recorded only under a broad human or service account, investigation may not reveal whether a person, agent, or integration caused a change. The CSA–Strata finding that fewer than a third of respondents could reliably trace actions across all environments highlights this attribution problem. The CSA–Zenity response-time and detection-confidence results further suggest that organizations need alerting and containment procedures designed around agent behavior, not just conventional user activity.
Lifecycle and retirement
Agents need a defined end-of-life process as well as an onboarding process. When a pilot ends, a team changes, or a workflow is replaced, the organization should be able to disable the agent, revoke credentials and permissions, remove integrations, and preserve records needed for accountability. The CSA–Token survey found that 21% of respondents had a formal agent decommissioning process, indicating that retirement is often less established than deployment.
Rank #3
- POWERFUL SECURITY KEY: The YubiKey 5C NFC is the most versatile physical passkey, protecting your digital life from phishing attacks. It ensures only you can access your accounts
- WORKS WITH 1000+ ACCOUNTS: Compatible with popular accounts like Google, Microsoft, and Apple. A single YubiKey 5C NFC secures 100+ of your favorite accounts, including email, password managers, and more
- FAST & CONVENIENT LOGIN: Plug in your YubiKey 5C NFC via USB and tap it, or tap it against your phone (NFC), to authenticate. No batteries, no internet connection, and no extra fees required
- MOST SECURE PASSKEY: Supports FIDO2/WebAuthn, FIDO U2F, Yubico OTP, OATH-TOTP/HOTP, Smart card (PIV), and OpenPGP. That means it’s versatile, working almost anywhere you need it
- PRIMARY & SPARE KEYS: Just like having a spare house key, we recommend buying two YubiKeys - one for daily use and one as a spare. That way you’ll never get locked out of your accounts
A practical AI agent readiness review
Use these questions to assess the integration around each agent or agentic workflow. They are control themes reflected in the surveys, not a statement of legal requirements.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →- Discover: Can teams find agents across relevant SaaS, cloud, endpoint, and development environments, including unsanctioned or embedded agents?
- Assign ownership: Is there a named business owner and technical owner, with an approved purpose and a way to review changes?
- Limit access: Which data, applications, and tools can the agent reach? Are permissions restricted to the task, and is authorization checked at the point of action?
- Control credentials: What identity and credentials does the agent use? Can credentials be scoped, rotated, revoked, and distinguished from a person’s own credentials?
- Trace actions: Can logs attribute each consequential action to the agent and the responsible person or system, including the resources accessed and changes made?
- Detect and contain: Can security teams identify unusual or unauthorized behavior, stop the agent, and revoke its access without waiting for a full incident investigation?
- Retire safely: When the workflow ends, is there a documented way to disable the agent, revoke access, remove integrations, and retain necessary audit records?
For each answer, record the system or team responsible and any gap that needs remediation. This turns a general concern about agent security into specific integration work that can be assigned and revisited as the agent’s purpose or access changes.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What the evidence can—and cannot—establish
CSA’s 2025 State of AI Security and Governance resource, commissioned by Google, identifies governance maturity, staff skills gaps, and data exposure as enterprise readiness themes. Its available summary does not provide enough methodology or specific figures to support quantitative conclusions.
Rank #4
- POWERFUL SECURITY KEY: The Security Key NFC is the essential physical passkey for protecting your digital life from phishing attacks. It ensures only you can access your accounts.
- WORKS WITH 1000+ ACCOUNTS: Compatible with Google, Microsoft, and Apple. A single Security Key NFC secures 100 of your favorite accounts, including email, password managers, and more.
- FAST & CONVENIENT LOGIN: Plug in your Security Key NFC via USB-A and tap it, or tap it against your phone (NFC) to authenticate. No batteries, no internet connection, and no extra fees required.
- TRUSTED PASSKEY TECHNOLOGY: Uses the latest passkey standards (FIDO2/WebAuthn & FIDO U2F) but does not support One-Time Passwords. For complex needs, check out the YubiKey 5 Series.
- BUILT TO LAST: Made from tough, waterproof, and crush-resistant materials. Manufactured in Sweden and programmed in the USA with the highest security standards.
A separate 2026 NeuralTrust report summary says 72% of surveyed organizations had deployed or were scaling agents and 29% had comprehensive controls. Its incident finding is based on 164 respondents. This is a distinct vendor-produced sample, not a measure that can be pooled with Gartner’s or CSA’s results.
Across these sources, the evidence supports a synthesis: reported agent use is widespread in some surveyed populations, while respondents also describe weaknesses in visibility, governance, permissions, attribution, response, and retirement. It does not establish a single causal estimate of harm, a universal rate of agent incidents, or that every organization has the same readiness gap.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
