How should security teams reduce the risk of MFA fatigue attacks?

Why This Matters for Security Teams

MFA fatigue works because attackers turn user-facing approval prompts into a denial-of-security problem. The issue is not just repeated notifications, but the assumption that a human will eventually approve one. For high-risk access paths, that assumption is now too weak. Current guidance suggests replacing approval-based MFA with cryptographic authentication and reducing the value of any single approved session through tighter privilege boundaries and stronger session controls.

NHIMG research shows how quickly identity abuse can become active compromise: in The 52 NHI breaches Report, credential failures repeatedly appear as a root cause class, reinforcing that identity controls fail fastest when they depend on human tolerance rather than machine-enforced proof. That pattern is consistent with broader defensive guidance in the NIST Cybersecurity Framework 2.0, which pushes organisations toward stronger access control, monitoring, and response discipline rather than alert fatigue.

For teams protecting sensitive admin paths, cloud consoles, and privileged support workflows, the practical objective is to make one successful prompt less useful and repeated prompting more suspicious. In practice, many security teams encounter MFA fatigue only after a user has already been conditioned to approve prompts, rather than through intentional detection design.

How It Works in Practice

The most effective response is to remove approval fatigue from the attack surface entirely. Where the access path is privileged or highly sensitive, use phishing-resistant authentication, step-up checks tied to cryptographic proof, and session scoping that limits what a successful login can do. For NHI-heavy environments, this pairs well with Top 10 NHI Issues, which highlights how credential misuse and overreach become systemic when identity is treated as a one-time gate instead of an ongoing control point.

Operationally, security teams should treat repeated prompts as a signal, not noise. That means correlating prompt bursts with geography, device posture, impossible travel, new token issuance, and unusual access timing. It also means using response playbooks that can pause the account, revoke sessions, and force re-verification before an attacker can turn annoyance into access. Where available, platform telemetry should be routed into detection rules that distinguish one-off user error from coordinated abuse. The Anthropic — first AI-orchestrated cyber espionage campaign report is useful here because it shows how automation can amplify adversary speed and persistence, which makes rapid prompt-led escalation more dangerous.

Where identity governance is mature, teams also shrink standing privilege, apply JIT elevation, and ensure that a completed session does not preserve unnecessary reach. That design matters because MFA fatigue is often paired with post-authentication abuse: the attacker does not need endless prompts if the first approved session is too powerful. These controls tend to break down when legacy VPNs, shared admin accounts, or flat RBAC models still grant broad access after a single approval.

Common Variations and Edge Cases

Tighter access control often increases friction for administrators and service desk teams, so organisations must balance usability against the blast radius of a compromised approval. Best practice is evolving, but there is no universal standard for this yet: some environments can remove push-based MFA entirely for privileged roles, while others need layered controls during transition. The key is to avoid treating all login contexts the same.

High-volume support desks, break-glass accounts, and contractor access are common edge cases. Those paths often need separate handling because prompt fatigue can be confused with normal operational activity. In these cases, response logic should be tuned around frequency, privilege, and time window rather than raw notification count alone. For broader context on identity abuse patterns, see Microsoft Midnight Blizzard breach and Anthropic — first AI-orchestrated cyber espionage campaign report, both of which underscore how identity abuse accelerates once attackers find a reliable entry path.

For teams modernising their control stack, the practical rule is simple: if approval prompts are still part of the trust model, make them rare, high-signal, and non-authoritative for privileged actions. If they remain common, attackers will keep using them as a pressure point rather than a barrier.

Related resources from NHI Mgmt Group

• How should teams reduce the risk of exposed AI credentials being abused?
• How should teams reduce risk from malicious npm package installs?
• How should security teams reduce the risk of secret theft from npm supply chain attacks?
• How should security teams reduce the risk of SSO bypass attacks?