A slow-and-low attack is a deliberate, low-volume intrusion pattern designed to avoid detection. Attackers spread actions over time, keep requests close to normal behavior, and use small deviations instead of obvious spikes. This makes conventional anomaly detection less effective and increases the value of contextual, sequence-based analysis.
Expanded Definition
A slow-and-low attack is not defined by a single tool or payload, but by the attacker’s tempo. In NHI security, that usually means stolen API keys, service account tokens, or agent credentials are exercised in small bursts that resemble ordinary automation, making rate thresholds and volume-only alerts less effective. The pattern is closer to patient tradecraft than brute force: the actor preserves access, avoids noisy failures, and lets each action blend into baseline traffic. That aligns with how sequence-aware monitoring is described in MITRE ATT&CK Enterprise Matrix and why NHI governance guidance from Ultimate Guide to NHIs — Why NHI Security Matters Now emphasizes visibility over mere credential presence. Guidance varies across vendors on whether the term should include purely reconnaissance activity or only abuse that culminates in persistence, so usage in the industry is still evolving. The most common misapplication is treating any low-volume traffic as benign, which occurs when defenders rely on static thresholds without context, identity lineage, or request sequencing.
Examples and Use Cases
Implementing defenses against slow-and-low activity rigorously often introduces monitoring overhead, requiring organisations to weigh richer behavioral visibility against alert fatigue and storage cost.
- A compromised API key is used once every few hours to enumerate buckets or endpoints, staying below volume-based anomaly thresholds while still advancing recon.
- An AI agent token is replayed intermittently to extract small responses from a model-backed workflow, matching normal cadence closely enough to avoid a burst alert. This threat pattern is consistent with the kind of credential abuse discussed in LLMjacking: How Attackers Hijack AI Using Compromised NHIs.
- A service account with excessive privilege performs a few legitimate-looking reads per day, then slowly escalates access after weeks of inactivity. Attack sequencing matters here, as reflected in MITRE ATT&CK Enterprise Matrix.
- Attackers test exposed cloud credentials at a measured pace to blend into normal failed-authentication noise, especially when secrets are found in code or CI/CD pipelines. NHIMG’s Ultimate Guide to NHIs — Key Challenges and Risks shows why these hidden credentials create durable exposure.
- A malicious agent changes its behavior slightly across runs, using small variations in prompts, destinations, or timing to evade simple signature rules.
Why It Matters in NHI Security
Slow-and-low attacks are dangerous because NHI environments often contain long-lived credentials, broad entitlements, and machine-to-machine traffic that already looks repetitive. That combination gives an attacker room to move quietly for days or weeks, especially when secrets are exposed outside managed vaults or service accounts are rarely reviewed. NHIMG reports that only 5.7% of organisations have full visibility into their service accounts, which makes tempo-based abuse much harder to spot early. The issue is not just detection, but governance: if a token remains valid, overprivileged, and untracked, low-and-slow abuse can persist long after the initial compromise. The Top 10 NHI Issues and the 52 NHI Breaches Analysis both reinforce that identity-centric visibility is essential when compromise does not announce itself with volume. Organisations typically encounter the true cost only after data exfiltration, privilege expansion, or downstream agent misuse is discovered, at which point slow-and-low detection becomes 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 AI RMF set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Non-Human Identity Top 10 | NHI-02 | Slow credential abuse is a core NHI detection and secret misuse concern. |
| OWASP Agentic AI Top 10 | A9 | Agentic systems can be abused through subtle, low-tempo action patterns. |
| NIST CSF 2.0 | DE.CM-1 | Continuous monitoring is required to catch low-and-slow intrusion patterns. |
| NIST Zero Trust (SP 800-207) | Zero Trust assumes continuous verification of every request and session. | |
| NIST AI RMF | AI risk management must address subtle abuse of models and agents over time. |
Assess iterative misuse risks and build controls that detect gradual degradation or manipulation.
Related resources from NHI Mgmt Group
- Why do low-severity or long-standing bugs become more dangerous in AI-assisted attack scenarios?
- Why do fixed traffic rules miss low-and-slow attacks?
- What do security teams get wrong about low-and-slow application probing?
- What signals show that identity response is too slow for modern attack pacing?
Deepen Your Knowledge
Reviewed and updated by the NHIMG editorial team on August 27, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org