Not every integrity incident is cheating. Some are delivery failures, marking-platform outages, or vendor problems that damage result trust even when no candidate has broken the rules. The sources show why assessment leaders need to separate misconduct from operational failure, because the right response is different: one needs investigation and sanction, the other needs containment, correction, and recovery;;.

Delivery and platform failure covers glitches, outages, marking-system weaknesses, and security vulnerabilities that affect the administration, scoring, or handling of an assessment. Misconduct covers cheating, impersonation, leakage, tampering, or unauthorised assistance by candidates or insiders;.

These failures often get grouped together in public debate, but they call for different operational decisions. A security breach in a marking platform can invalidate trust in scores without implying candidate wrongdoing, while cheating may require discipline, retest decisions, or sanctions. For assessment leaders, the practical issue is speed and classification. If the programme cannot quickly tell which kind of failure has happened, it risks overreacting in one direction and underreacting in another.

- **Delivery failure**: a problem in the process of running or delivering the assessment. - **Platform failure**: a technical or security issue in the system used to mark, host, or process results. - **Misconduct**: deliberate rule-breaking by candidates, staff, or other actors. - **Recovery**: the route for correction, rerun, retake, or communication after a failure.

The sources point to a shared operational principle: assessment systems need both prevention and recovery. A suspended marking platform and a glitch-ridden bar exam both show that trust can be damaged by technical failures that are not classic cheating incidents;. There is also a strong governance lesson from the California case: settlement may resolve a dispute, but it does not answer the programme-design question of how to prevent repeat failures.

The contest is over where responsibility sits and how far the vendor should carry it. Some failures are clearly supplier-side, but the assessment owner still has to decide how to protect candidates and whether the delivery model remains defensible. There is also an open question about when a security vulnerability in a marking platform is best treated as a cyber incident, a quality incident, or both. The practical answer is often both.

- confusing technical failure with candidate misconduct - delaying corrections while blame is sorted out - underestimating the impact of platform vulnerabilities on result trust - treating settlement or apology as the end of the matter - failing to preserve evidence and logs for review

1. Classify the incident quickly. - Decide whether the issue is misconduct, platform failure, delivery failure, or a mix. 2. Contain the problem. - Freeze affected workflows, preserve logs, and stop further harm where possible. 3. Protect unaffected candidates. - Avoid broad remedies that penalise people outside the affected cohort. 4. Choose the right recovery route. - Use score correction, retest, alternative evidence, or platform replacement as needed. 5. Review the operating model. - Ask whether the same failure could recur and what governance or vendor controls need strengthening.

| Response | Strength | Limitation | Best fit | |---|---|---|---| | **Contain and correct** | Fastest way to protect fairness after a technical failure | Does not remove the underlying cause | Scoring glitches, platform bugs, data corruption | | **Investigate and sanction** | Appropriate when there is deliberate misconduct | Can be too slow if the problem is technical | Cheating, leakage, tampering, impersonation | | **Redesign delivery** | Reduces recurring platform risk | Requires change management and revalidation | Repeated outages or security vulnerabilities |

A national exam board suspends its marking platform after security failings are identified. The immediate priority is not to ask which candidate cheated, but to secure the system, check which results may be affected, and decide how to restore confidence in the marking workflow. A separate bar-exam case ends in settlement after repeated glitches. That kind of outcome may resolve the dispute, but the deeper reader question is whether the same delivery model should be trusted again without stronger controls.

- News note on a national marking platform suspension after security failings. - News note on a lawsuit settlement after bar-exam glitches. - Discussion of integrity failures, response, and remediation.

This comparison sits in the vendor and platform risk space. Buyers should ask how a supplier separates operational fault handling from security incident handling, and whether the contract covers correction, communication, and recovery rather than only uptime.

### Is a platform glitch the same as cheating? No. It is a different failure mode, even though both damage trust. ### What should I ask first after a marking-platform incident? Ask whether the problem is technical failure, security compromise, or both, and which results may be affected. ### Can a settlement restore trust? It can help resolve the dispute, but trust usually depends on transparency, remediation, and evidence that the failure will not repeat.

Tim Burnett (Admin)

`Test Community Network. "Delivery and platform failure versus misconduct in assessment security." TCN AI & Assessment Wiki. Last reviewed 2026-08-12. https://testcommunity.network/knowledge/test-security/wiki/test-security-detection-vs-online-platform-failure`

- News note on a national marking platform suspension after security failings. - News note on a lawsuit settlement after bar-exam glitches. - Discussion of integrity failures, response, and remediation.