Audits human oversight requirements per Article 14 of the EU AI Act. Checks for human-in-the-loop, human-on-the-loop, and human-in-command mechanisms, override capabilities, and kill switch functionality for high-risk AI systems.
Audit compliance with human oversight requirements for high-risk AI systems.
Required: High-risk AI systems shall be designed and developed to be effectively overseen by natural persons during the period of use.
Search patterns:
human.?oversight|human.?control|human.?supervision
oversight.?design|oversight.?mechanism|oversight.?interface
operator|supervisor|reviewer|moderator
Required: Human oversight shall aim to prevent or minimize risks to health, safety, or fundamental rights, particularly automation bias.
Search patterns:
automation.?bias|over.?reliance|over.?trust
confirmation.?bias|anchoring.?bias|decision.?support
advisory.?only|recommendation.?only|not.?deterministic
human.?judgment|human.?decision|human.?final
calibration.?warning|confidence.?display|uncertainty
Evidence to check:
Required: Persons overseeing the system must be able to fully understand the system's capabilities and limitations, and properly monitor its operation.
Search patterns:
operator.?training|user.?training|oversight.?training
capability.?document|limitation.?document
monitoring.?dashboard|operator.?interface|admin.?panel
system.?status|health.?indicator|performance.?display
Required: Overseers must remain aware of the possible tendency of automatically relying on or over-relying on the output.
Search patterns:
automation.?bias.?warning|reliance.?warning
training.*automation.?bias|awareness.*over.?reliance
decision.?fatigue|alert.?fatigue|monitoring.?fatigue
Required: Overseers must be able to correctly interpret the system's output, taking into account the tools and methods of interpretation.
Search patterns:
interpret.*output|output.*explain|result.*interpret
visualization|dashboard|report.?generat
explainab|SHAP|LIME|feature.?importance
decision.?rationale|reasoning.?display|explain.?why
Required: Overseers must be able to decide not to use the system or to disregard, override, or reverse its output.
Search patterns:
override|disregard|reverse.?decision|reject.?output
manual.?override|human.?override|operator.?override
opt.?out|decline|dismiss|ignore.?recommendation
veto|countermand|overrule|bypass
Evidence to check:
Required: Overseers must be able to interrupt or stop the system through a "stop" button or similar procedure.
Search patterns:
stop.?button|kill.?switch|emergency.?stop|e.?stop
interrupt|halt|shutdown|abort|terminate
circuit.?breaker|dead.?man|watchdog
graceful.?shutdown|safe.?stop|pause
Evidence to check:
Required: For high-risk AI systems used for biometric identification, human oversight ensures:
Search patterns:
dual.?review|two.?person|four.?eyes|independent.?verif
human.?confirm|manual.?confirm|second.?opinion
geographic.?limit|temporal.?limit|scope.?restrict
verification.?required|confirmation.?required
Required: Provider shall identify appropriate human oversight measures and include them in instructions for use.
Search patterns:
oversight.?instruction|oversight.?guide|oversight.?manual
deployer.?guide|operator.?manual|user.?guide
human.?oversight.?section|oversight.?procedure
Evaluate the level of human oversight:
| Level | Description | Pattern |
|---|---|---|
| HITL (Human-in-the-Loop) | Human approves every decision | approval.?required|manual.?confirm|queue.?review |
| HOTL (Human-on-the-Loop) | Human monitors and can intervene | monitor|dashboard|alert|override |
| HICC (Human-in-Command) | Human sets boundaries, AI operates within | policy|constraint|boundary|rule.?engine |
## Human Oversight Audit (Article 14)
| Requirement | Article | Status | Evidence |
|-------------|---------|--------|----------|
| Designed for oversight | 14(1) | [STATUS] | [evidence] |
| Automation bias prevention | 14(2) | [STATUS] | [evidence] |
| Understand capabilities | 14(3)(a) | [STATUS] | [evidence] |
| Automation bias awareness | 14(3)(b) | [STATUS] | [evidence] |
| Interpret output | 14(3)(c) | [STATUS] | [evidence] |
| Override/disregard | 14(3)(d) | [STATUS] | [evidence] |
| Interrupt/stop | 14(3)(e) | [STATUS] | [evidence] |
| Identification oversight | 14(4) | [STATUS] | [evidence] |
| Provider documentation | 14(5) | [STATUS] | [evidence] |
**Oversight Architecture**: [HITL / HOTL / HICC / NONE]
**Overall Art. 14 Compliance**: [PERCENTAGE]%
### Critical Gaps
[List most important missing elements]
### Remediation Priority
1. [Highest priority action]
2. [Second priority]
3. [Third priority]