TL;DR: AI-enabled attacks surged 1210% in 2025, with non-AI fraud also rising 195% at year end, while healthcare saw more than 15,000 unique bot calls and retailers recorded a 330% increase in AI fraud in two months, according to Pindrop research. The scale shift means fraud controls must move from static detection to channel-aware, adaptive identity governance.
At a glance
What this is: Pindrop says AI-enabled fraud is accelerating across voice, bot, and refund abuse channels, with 2025 growth far outpacing non-AI fraud.
Why it matters: Fraud now spans customer identity, support workflows, and automated channel abuse, so IAM and fraud teams need controls that treat identity signals as dynamic rather than one-time checks.
By the numbers:
- AI-enabled attacks surged 1210% in 2025, according to Pindrop.
- Non-AI fraud also increased 195% at the end of 2025, according to Pindrop.
- Among major retailers and Pindrop customers, AI fraud increased 330% in two months, according to Pindrop.
Context
AI fraud is the use of automation, generative models, and synthetic media to scale deception across voice, chat, and account workflows. In practice, it turns one-off fraud attempts into repeatable operations that can adapt to controls faster than traditional review processes can keep up.
The main governance gap is that many organisations still treat fraud and identity verification as static checkpoint problems. That assumption breaks when attackers can imitate customers, overwhelm support queues, or automate refund abuse at machine speed.
Key questions
Q: How should security teams respond when AI fraud starts overwhelming customer and support channels?
A: They should treat the problem as a workflow integrity issue, not only a detection issue. Separate channels by risk, step up verification for recovery and payout actions, and correlate identity signals with bot telemetry, session behaviour, and support context. The goal is to stop attackers from turning one successful impersonation into repeatable abuse across many accounts.
Q: Why do AI-generated voice and synthetic media make account recovery more dangerous?
A: Because recovery workflows often accept trust once the claimant sounds or looks legitimate. AI-generated media can supply just enough credibility to trigger resets, payment changes, or refund approvals. Without stronger re-authentication and out-of-band validation, recovery becomes a privileged path that attackers can reuse at scale.
Q: What are the signs that a fraud program is falling behind bot-driven abuse and account takeover?
A: A program is falling behind when bot traffic rises, customer support becomes a frequent recovery path for compromised accounts, and fraud moves into peripheral channels such as loyalty, referrals, or discounts. Those signals show attackers are adapting faster than controls, and that the organisation needs stronger detection across the full customer journey.
Q: Should organisations prioritise bot detection or identity proofing first for AI fraud?
A: They should do both, but prioritise the workflows where identity evidence changes account state. Bot detection helps spot volume and automation, while stronger identity proofing reduces synthetic enrolment and recovery abuse. If a process can alter access, payment, or ownership, it needs both controls working together.
Technical breakdown
How AI changes fraud economics across channels
AI does not create new fraud objectives so much as it collapses the cost of repetition, personalization, and adaptation. A fraudster can test prompts, generate convincing voice or video, and iterate on scripts until a channel yields. That matters because fraud detection models and manual review queues were built for slower adversaries with higher operating costs. When the attacker can automate the entire social engineering loop, volume becomes a weapon. The result is not only more attempts, but a faster feedback cycle that degrades legacy rules and thresholds. Practical implication: move from single-point verification to layered, signal-rich fraud controls that can adapt per channel and per session. See the Identity Fraud Prevention Guide.
Practical implication: Shift fraud controls from static checkpoints to layered, channel-aware detection that can respond in-session.
Why support and recovery flows become identity attack paths
Customer support, account recovery, and refund handling are identity-sensitive workflows because they often bypass normal login friction once the caller or claimant seems credible. AI-generated speech and synthetic personas make those paths easier to exploit at scale. The issue is not just impersonation, but trust transference: once a support process accepts the premise that a caller is legitimate, downstream actions can include password resets, payment redirection, or refund issuance. In identity terms, these flows become delegated verification points without sufficient re-authentication depth. Practical implication: treat support and recovery as privileged identity workflows, not as back-office operations. See the Customer IAM (CIAM) Guide.
Practical implication: Apply step-up verification and fraud telemetry to support, recovery, and refund actions as privileged identity events.
What bot surges reveal about automated abuse
Large bot spikes show that fraud is no longer confined to human-led social engineering. Automated traffic can probe rate limits, session handling, and workflow weak points until it finds the cheapest path to abuse. The fact that one healthcare provider saw more than 15,000 unique bot calls is important because it shows how quickly a service can be saturated even before any single account is taken over. These campaigns also blur the line between fraud, abuse, and availability pressure, because volume itself becomes the mechanism of attack. Practical implication: instrument bot detection as a fraud control, not only as an availability safeguard.
Practical implication: Use bot telemetry as part of fraud governance, because volume attacks can precede account abuse and service disruption.
Threat narrative
Attacker objective: The attacker wants to convert synthetic credibility into repeatable financial fraud, account takeover, or workflow abuse at scale.
- Entry occurs through AI-generated outreach, synthetic voice, or automated bot traffic that looks credible enough to reach the target workflow.
- Credential or trust capture follows when the victim or support process accepts the impersonation and discloses access, approves recovery, or processes a refund.
- Escalation and impact occur as the attacker repeats the flow at scale, converting one successful interaction into many account takeovers, fraudulent refunds, or social engineering wins.
Breaches seen in the wild
- Gitloker GitHub extortion campaign: Phishing via GitHub notifications tricked developers into authorising malicious OAuth apps; Gitloker then wiped repos and demanded contact.
- 23andMe credential stuffing 2023: Credential stuffing into 18,000 accounts exposed nearly 7 million people through the DNA Relatives feature.
Read and download The State of NHI & AI Agent Breach Report 2026, covering 200+ breaches impacting Non-Human Identities including AI Agents.
NHI Mgmt Group analysis
AI fraud is now a governance problem, not just a detection problem. Once fraudsters can use synthetic media and automation to scale personalisation, the issue moves beyond fraud rules into identity governance, channel assurance, and recovery control. The control question is no longer whether a user can pass one check, but whether the workflow can resist repeated, adaptive abuse. Practitioners should treat AI fraud as a cross-functional identity risk, not a single-team alerting issue.
Customer recovery and support flows are the new privileged paths. AI fraud turns assistance channels into trust transfer mechanisms, especially where agents or automation can approve resets, refunds, or account changes. That means the governance boundary must extend beyond login to every place where identity evidence is accepted and acted on. The practical conclusion is straightforward: if a workflow can change account state, it deserves privileged treatment.
Fraud programmes need a named concept for synthetic channel pressure. Synthetic channel pressure: the point at which AI-generated traffic, impersonation, and bot volume overwhelm a channel faster than manual review or static rules can respond. This is not a niche edge case; it is the operating model attackers now prefer because it scales cheaply and degrades confidence in ordinary verification. Practitioners should design controls around channel integrity, not just user identity.
Identity verification alone will not absorb this threat class. The article shows that fraud is spreading across voice, bots, and refund processes, which means isolated IDV checks leave large operational gaps. The broader lesson is that fraud defence must be continuous, contextual, and linked to account lifecycle signals. Teams that still rely on one-time verification will keep finding that the attacker moved faster than the trust decision.
What this signals
Synthetic channel pressure: AI fraud changes the attacker model from isolated deception to high-volume, adaptive pressure on support and verification channels. That means the defensive unit is no longer the single login event, but the end-to-end workflow that can be induced to trust the wrong party.
Fraud teams should expect more blended attacks that combine bot traffic, synthetic media, and recovery abuse in one campaign. The organisations that fare better will be the ones that instrument identity, session, and workflow signals together instead of treating fraud controls as separate point solutions.
For practitioners
- Build channel-specific fraud controls Separate voice, chatbot, web, and recovery workflows in policy so each channel gets different risk scoring, step-up checks, and response thresholds.
- Treat recovery as a privileged workflow Require stronger re-authentication, out-of-band confirmation, and evidence logging before password resets, refund issuance, or account changes.
- Instrument bot traffic as a fraud signal Correlate bot volume, request velocity, and repeated failed attempts with account state changes and support interactions.
- Harden identity proofing against synthetic media Use document, liveness, and anomaly checks that can resist AI-generated voice, image, and video artefacts during onboarding and recovery.
Key takeaways
- AI fraud now exploits the same business workflows that resolve accounts, issue refunds, and reset access, which makes channel governance as important as detection.
- The article shows how scale changes the problem, with bot surges and rapid growth rates indicating that attackers can outpace manual review and static rules.
- Practitioners should strengthen recovery, proofing, and bot telemetry together so identity assurance extends to every action that can change account state.
Related NHI Mgmt Group resources: Identity Fraud Prevention Guide · Customer IAM (CIAM) Guide · Identity Proofing and KYC Guide · Account Takeover Prevention Guide
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 addresses the attack and risk surface, while NIST SP 800-63, NIST CSF 2.0 and CIS Controls v8 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Non-Human Identity Top 10 | NHI-04 — Insecure Authentication | AI fraud exploits weak identity checks in recovery and support workflows. |
| NHI-10 — Human Use of NHI | Fraudsters use automation and synthetic identity signals to impersonate legitimate users at scale. | |
| Recommendation — Strengthen authentication at recovery and support touchpoints to prevent synthetic impersonation from changing account state. Separate human-led from automated interactions and enforce additional checks where humans and bots share the same workflow. | ||
| NIST SP 800-63 | SP 800-63B — Authentication | The article centres on authentication strength in high-risk fraud and recovery flows. |
| Recommendation — Apply stronger authentication and re-authentication controls to workflows that can reset access or move funds. | ||
| NIST CSF 2.0 | PR.AA-05 — Access Permissions, Entitlements and Authorizations | Fraudulent support actions can alter account permissions and entitlements. |
| Recommendation — Restrict state-changing account actions to authorised workflows with explicit approval and logging. | ||
| CIS Controls v8 | CIS-5 — Account Management | Account takeover and recovery abuse are central to the article's fraud pattern. |
| Recommendation — Monitor account changes, recovery events, and anomalous access patterns for signs of fraud-driven abuse. | ||
Key terms
- Synthetic Channel Pressure: The concentration of AI-generated calls, messages, and automated requests on a business channel until manual review and static rules become ineffective. It is a governance problem as much as a detection problem because the attacker is targeting the trust boundary of the workflow itself.
- Recovery Abuse: Recovery abuse is the use of password reset, fallback verification, or support escalation paths to take over an account after the primary login is defended. It matters because many programmes harden the front door while leaving the back door easier to open under social engineering or session manipulation.
- Bot-Driven Fraud: Fraud activity amplified by automation that allows attackers to repeat attempts, vary inputs, and scale abuse across many accounts. The identity risk is not only volume but adaptability, because bots can keep probing until they find a path past controls designed for individual users.
- Identity proofing: The process of verifying that a person is who they claim to be before granting or restoring access. In higher-risk recovery paths, proofing can include stronger evidence checks such as government ID validation or liveness-based facial verification so the assurance level matches the sensitivity of the request.
Deepen your knowledge
NHI Mgmt Group covers identity security, NHI governance, and agentic AI through independent research, practitioner guides, and the NHI Foundation Level course, the industry's only accredited NHI security programme. It is designed for practitioners who need to connect identity controls to broader security and fraud programmes.
Published by the NHIMG editorial team on October 22, 2026.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org