Voice phishing that uses synthetic or cloned speech to impersonate a trusted person and pressure a target into approving an identity action. In practice, it is a social engineering attack against recovery and support workflows, not just against end users.
Expanded Definition
AI vishing is a voice phishing technique that uses synthetic or cloned speech to sound like a trusted executive, colleague, help desk agent, or family member. In NHI security, the target is usually not a password alone but an identity workflow, such as account recovery, MFA reset, privileged approval, or delegated access. The attack works because the victim is persuaded that the caller has authority and urgency, then makes an identity action that bypasses normal verification.
Definitions vary across vendors on whether AI vishing is a separate category or simply an enhanced form of social engineering, but the operational issue is clear: it combines impersonation, urgency, and channel trust. This makes it especially dangerous in environments where recovery procedures rely on human discretion instead of strong proof of identity. NIST’s NIST Cybersecurity Framework 2.0 is useful here because it frames identity assurance as part of broader governance and protective controls, not just endpoint security. The most common misapplication is treating AI vishing as an ordinary spam call, which occurs when teams ignore voice-driven approval paths in recovery and support workflows.
Examples and Use Cases
Implementing strong defences against AI vishing often introduces friction in help desk and executive support processes, requiring organisations to weigh faster recovery against stronger identity proofing.
- A cloned CFO voice calls finance support and pressures an agent to approve a password reset for a payroll system.
- A synthetic manager voice convinces a service desk to bypass normal step-up checks during a locked-account recovery.
- An attacker uses a convincing voice note to trigger out-of-band approval from a delegated approver in an identity governance workflow.
- A threat actor exploits a callback process by spoofing a known employee and steering staff toward a fraudulent reset path, similar in style to the social-engineering chains discussed in the MGM Resorts Breach 2023 — Scattered Spider analysis.
- A company updates recovery controls after reviewing lessons from the Caesars Entertainment Breach 2023 — Scattered Spider incident and requiring stronger verification before privileged changes.
For implementation patterns around identity-centered defence, the NIST Cybersecurity Framework 2.0 supports control design that ties recovery actions to validated trust signals rather than caller confidence alone. AI vishing also intersects with synthetic media abuse seen in the DeepSeek breach coverage, where security failures show how quickly attacker access can escalate once trust is misplaced.
Why It Matters in NHI Security
AI vishing matters because it turns human trust into an identity exploit path. In NHI environments, that means a forged voice can become the entry point for secret exposure, token reset, privileged elevation, or support-channel abuse. Once an attacker convinces a human operator to approve an action, the technical controls around NHI, secrets, and delegated access may be bypassed entirely. That is why voice-driven impersonation should be treated as an identity assurance problem, not only as fraud prevention.
NHIMG research on secrets management shows why this risk compounds quickly: only 44% of developers are reported to follow security best practices for secrets management, and the average estimated time to remediate a leaked secret is 27 days, giving attackers a long window after one successful social-engineering event. The State of Secrets in AppSec findings reinforce that weak human and workflow controls can make a single approval mistake disproportionately costly. Organisations typically encounter the operational reality of AI vishing only after a reset, transfer, or approval has already been abused, at which point recovery workflows become operationally unavoidable to address.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP Non-Human Identity Top 10 and OWASP Agentic AI Top 10 address the attack and risk surface, while NIST CSF 2.0, NIST Zero Trust (SP 800-207) and NIST SP 800-63 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Non-Human Identity Top 10 | NHI-05 | Covers identity recovery and approval abuse that voice impersonation tries to exploit. |
| OWASP Agentic AI Top 10 | A-03 | Addresses social manipulation of autonomous or assisted decision workflows. |
| NIST CSF 2.0 | PR.AC-7 | Identity proofing and access approval discipline align with controlled access outcomes. |
| NIST Zero Trust (SP 800-207) | SP 800-207 | Zero trust rejects trust based on channel familiarity or caller persona. |
| NIST SP 800-63 | IAL2 | Identity proofing strength informs how much trust a recovery action should receive. |
Gate high-risk actions behind independent verification before agents or humans execute them.
Related resources from NHI Mgmt Group
Deepen Your Knowledge
Reviewed and updated by the NHIMG editorial team on August 11, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org
