Arkose Labs’ banking fraud data exposes AI attack readiness gaps

At a glance

What this is: This is Arkose Labs' banking-focused analysis showing that concern about account takeover and credential stuffing is high, but AI attack readiness and cross-skilled staffing remain low.

Why it matters: It matters because bank fraud controls, IAM governance, and AI threat response are now interdependent, so teams need to treat AI-enabled attack readiness as an identity and fraud operations problem, not a side issue.

By the numbers:

• 90% of banks are concerned about account takeover and credential stuffing.
• Only 19% of banks are very well prepared to combat AI attacks.
• 48% of banks lack personnel skilled in both AI and cybersecurity.

Context

Bank fraud teams are facing a governance gap: the threat is evolving toward AI-assisted attack patterns, but operational readiness still depends on legacy fraud controls, IAM processes, and scarce cross-skilled staff. In practice, that means the organisation may understand the risk without having the people, telemetry, and decisioning structure to act on it.

Credential stuffing, account takeover, fake account creation, and phishing all sit at the intersection of identity assurance and fraud operations. When AI increases the scale and variability of these attacks, the question shifts from whether banks recognise the threat to whether they can detect, decide, and mitigate fast enough across human, NHI, and agent-assisted abuse paths.

Key questions

Q: What breaks when banks rely on login controls to stop AI-assisted account takeover?

A: Login-only defence breaks when attackers can reuse valid credentials, vary their behaviour, and keep testing until one account path looks legitimate enough to pass. That leaves authentication as a thin gate rather than a risk decision. Banks need correlated device, behavioural, and identity signals so the control can distinguish human customers from automated abuse.

Q: Why do AI-related fraud threats matter to IAM teams?

A: AI-related abuse matters because it increases the speed and quality of deceptive interactions that target identity controls. IAM teams are affected when attackers can iterate social engineering, synthetic content, or automated abuse faster than static verification rules can react. That makes assurance, detection, and recovery controls part of the same risk model.

Q: How do security teams know whether AI data readiness is actually improving?

A: Look for shrinking exposure, not just more dashboards. The strongest signal is that sensitive data discovery leads to deletion, reclassification, or access reduction within defined service levels, and that AI requests increasingly hit correctly enforced policy boundaries instead of generating exceptions.

Q: Should banks treat AI-enabled account takeover as a fraud issue or an identity issue?

A: They should treat it as both. Account takeover is a fraud outcome, but the attack succeeds through identity weaknesses such as weak authentication, poor trust signals, and overreliance on reusable credentials. The strongest programme view is to govern it as a shared control problem across customer identity, fraud operations, and security response.

How it works in practice

Why AI attack readiness is a fraud operations problem

AI-assisted fraud changes the speed and shape of attack campaigns, but it does not remove the need for identity assurance. Banks still need to decide whether a session, account, device, or interaction is trustworthy, yet attack automation raises the volume and adaptability of signals they must evaluate. That creates pressure on risk scoring, step-up controls, and case handling. The technical issue is not simply detection accuracy. It is whether fraud controls can keep pace with adversaries that can vary timing, content, and infrastructure faster than manual review cycles.

Practical implication: align fraud decisioning, identity signals, and escalation logic so AI-amplified attacks are handled as a real-time control problem.

How credential stuffing and account takeover exploit identity assumptions

Credential stuffing succeeds when organisations assume reused credentials will be caught by login controls alone. Attackers exploit large credential sets, retry at scale, and look for authentication paths that lack enough context to distinguish legitimate users from automated abuse. Account takeover then follows when the attacker can satisfy weak or stale identity checks, often without needing to break cryptography or exploit code. In banking, the identity layer becomes the control surface. If risk scoring, device intelligence, and behavioural signals are not connected, the attacker can repeatedly test accounts until one behaves differently enough to matter.

Practical implication: treat reusable credentials and weak step-up logic as a combined control gap, not separate problems.

Why cross-skilled teams matter for AI and cybersecurity response

The article's staffing signal matters because AI attacks force fraud, IAM, and security teams to share the same operational language. Analysts need to understand identity risk, security engineers need to understand adversarial patterns, and response teams need to decide when to block, delay, or challenge a transaction or session. Without that overlap, investigations fragment across teams and remediation slows. This is a governance issue as much as a capability issue, because readiness depends on who can interpret the signal and who is authorised to act on it in the moment.

Practical implication: build joint fraud, IAM, and security playbooks so decision authority does not become the bottleneck.

NHI Mgmt Group analysis

Banking readiness for AI-driven fraud is now a control-plane problem, not an awareness problem. The article shows that concern about account takeover and credential stuffing is already widespread, but preparedness for AI attacks remains low. That gap means the issue is no longer recognition of threat category, but whether fraud, identity, and response controls can operate as one decisioning system. Practitioners should read this as a signal that operational alignment matters more than another point-in-time fraud policy.

Cross-skilled identity and cybersecurity teams are becoming a governance requirement. When 48% of banks lack personnel skilled in both AI and cybersecurity, the organisation is not just short of headcount. It is short of translation capacity between attack patterns, identity signals, and mitigation actions. That weakens escalation quality and delays containment. The implication is that staffing models for fraud and IAM can no longer be treated as separate planning exercises.

Identity trust in banking is being stretched by AI-assisted attack adaptation. Static authentication assumptions are brittle when adversaries can vary technique, timing, and scale around fixed controls. Banks that still depend on one-dimensional login defence are effectively asking legacy IAM to absorb a fraud problem that now includes machine-driven variation. The practitioner conclusion is that trust must be evaluated as a runtime condition, not a one-time authentication event.

Fraud mitigation and IAM now share the same evidence base. The strongest reading of this article is not that banks need more tools, but that they need a unified operational view of session risk, account behaviour, device trust, and escalation authority. That aligns with NIST CSF 2.0's focus on access permissions and risk governance, and with OWASP-NHI where machine-driven abuse becomes part of the identity attack surface. Teams should govern those signals together, or they will keep responding too late.

From our research library:

• 53% of security leaders expect AI to run major portions of their infrastructure autonomously within the next three years, according to the 2026 Infrastructure Identity Survey.
• 69% of security leaders agree identity management must fundamentally shift to address agentic AI systems, according to the 2026 Infrastructure Identity Survey.
• Read next: Top 10 Agentic AI Identity Issues

What this signals

Identity and fraud controls are converging: banks cannot separate account takeover defence from AI attack readiness anymore, because the same trust decisions now govern both human users and machine-amplified abuse. When identity signals are not shared across fraud and security workflows, response time becomes the weak link rather than detection.

The staffing gap is just as important as the tooling gap. If teams do not have people who can read both AI-shaped attack behaviour and banking identity controls, then the organisation will keep relying on fragmented escalation paths that slow containment and make automated abuse easier to repeat.

For practitioners

• Map AI-enabled fraud scenarios to shared identity controls Document how account takeover, credential stuffing, fake account creation, and phishing are handled across fraud, IAM, and security teams, then assign control ownership for each step of detection and mitigation.
• Strengthen risk-based step-up decisions Review where authentication, device reputation, and behavioural signals feed step-up or blocking decisions, and remove any path that lets high-volume automated abuse look like normal customer behaviour.
• Build joint fraud and IAM response playbooks Define who can challenge, suspend, investigate, or release an account when AI-assisted abuse is suspected, so response does not stall while teams negotiate ownership.
• Upskill analysts on AI attack patterns Create a cross-training path for fraud, security, and identity staff so they can recognise AI-shaped attack behaviour, interpret the telemetry, and use the same escalation language.

Key takeaways

• Banks are already concerned about account takeover and credential stuffing, but concern alone does not equal readiness when AI changes attack speed and variability.
• The article's most important evidence is the mismatch between 90% concern, 19% strong preparedness, and 48% cross-skilled staffing shortages.
• The practical response is to govern fraud, IAM, and security as one operating model so identity risk can be detected and mitigated in the same workflow.

Key terms

• Account Takeover: Account takeover is unauthorized use of a legitimate account after an attacker obtains valid access through stolen credentials, tokens, or trusted integrations. The key security problem is that the resulting activity often looks normal to logs and controls, which makes containment and attribution harder than in a forced-entry breach.
• Credential Stuffing: Credential stuffing is an attack that uses stolen username and password pairs from previous breaches to try logging into other services. It works because many people reuse credentials, and because the login attempt uses valid information, it can look ordinary until the surrounding behavior gives it away.
• AI-Enabled Fraud: AI-enabled fraud is deception carried out with the help of artificial intelligence to make scams faster, more convincing, or harder to detect. It includes synthetic identities, deepfakes, automated phishing, and adaptive social engineering. In security terms, it combines machine-generated content with identity abuse, fraud operations, and evasion of human and technical controls.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity lifecycle are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are building or maturing an IAM programme, it is worth exploring.