A survey commissioned by credential security vendor Dashlane reports that 72% of organizations experienced a security incident involving an AI agent's credentials or access in the past year. The study, conducted between August 18 and September 5, 2026, surveyed 3,000 full-time knowledge workers and 800 IT and security executives across the United States, United Kingdom, and France [1][2].
What the survey measured #
The research, titled "Misplaced Trust: Inside the Workplace's New AI Security Blind Spot," was conducted by Method Research and distributed by PureSpectrum. It used separate questionnaires for workers and security leaders, covering AI adoption, credential-sharing behavior, autonomous agent access, and security governance [1][5].
Dashlane, which sells the Omnix® platform for browser-based credential protection, framed the findings as evidence of a gap between employee trust in AI tools and the security controls organizations have in place [3][7].
Credential sharing and standing access #
Seventy-one percent of organizations reported granting AI agents standing access to internal systems or data. Twenty-one percent of workers admitted sharing a password, API key, or login credential with an AI tool or agent. Fifty-five percent of workers said they allow AI to act autonomously — including processing payments and accessing emails and customer data — without asking first [2].
The survey also found that 56% of workers granted AI privileges they would not extend to a human co-worker. Forty-eight percent said they would lend login credentials to an AI tool after a week of use, compared to 42% for a new human colleague. Fifty-two percent said they would comply if an AI asked them to confirm their login credentials daily [2][3].
Governance lagging deployment #
Fifty-three percent of IT and security leaders said some AI tools are deployed without a prior security review. Sixty-one percent identified pressure to accelerate AI adoption as one of the biggest barriers to securing agents. Among those facing such pressure, 57% said it resulted in faster deployment with less security review [2][5].
Ninety-three percent of security leaders reported feeling pressure from company leadership to adopt AI [3][7].
The visibility gap #
Eighty-nine percent of IT and security leaders said they were at least somewhat confident they knew about every AI agent in use. Yet 98% of organizations reported discovering unknown agents after deployment. Twenty-eight percent of workers admitted using AI tools their employer did not know about [2].
Fifty-four percent of security leaders said they were not fully confident they would hear about every AI-related incident in their organization [3][7].
Accountability and budget expectations #
When an incident occurs, 69% of IT and security leaders said responsibility would fall on them, while only 11% pointed to the business unit that deployed the agent. Eighty-four percent expect their budget for AI agent oversight to increase over the next 12 months, and 79% expect their oversight workload to grow [2][5].
Vendor interest and survey limitations #
Dashlane commissioned and funded the research. The company's Chief Product Officer, Christophe Frenet, said the stakes of getting AI security wrong are "much higher" than with previous technologies [2][5]. The findings align with Dashlane's product focus on browser-based credential protection and the Omnix platform [5].
The survey relies on self-reported behavior and perceptions. It does not include independent verification of incident rates, root causes, or the effectiveness of specific controls. The "72% incident" figure reflects respondents' answers, not a technical audit. No breakdown of incident severity, type, or attribution is provided in the released materials.
What remains unclear #
The report highlights an industry spotlight and country-level comparisons (U.S., U.K., France) that are not detailed in the public summary. The full report is gated behind a download form [3][7]. Independent studies replicating these findings have not been cited.
Sources #
- Survey Finds Widespread AI Credential Security Incidents (securitytoday.com)
- AI has the keys, and organizations are paying the price, Dashlane report finds (finance.yahoo.com)
3. [AI Is the New Co-Worker Employees Trust. That's The Problem.](https://www.dashlane.com/blog/ai-coworker-employee-trust) (dashlane.com)
4. [November 2026 Cyber Outages and Agentic AI](https://digitalmeld.io/blog/november-2026-cyber-outages-agentic-ai-defense) (digitalmeld.io)
6. [AI Agent Security Report: Misplaced Trust in Workplace AI | Dashlane](https://go.dashlane.com/report-misplaced-trust.html) (go.dashlane.com)
7. [AI Is the New Co-Worker Employees Trust. That's The Problem. | Dashlane](http://dashlane.com/blog/ai-coworker-employee-trust) (dashlane.com)