As high-stakes exams increasingly determine access to education, licensure, and professional opportunity, weaknesses in exam security carry significant downstream consequences. Compromised scores, stolen test content and delayed invalidations can erode trust and impose significant operational and reputational costs on testing programs.
In its recently completed a study, which spanned more than a year and included more than 100 remote and in-person exam administrations in the U.S. for professional certification and licensure exams, Caveon, the leading exam security provider in high-stakes testing, found exam proctors missed 90% of cheating attempts for both remote and in-person tests. And when violations were noticed, proctors most often issued a verbal warning and allowed testing to continue.
Among other key findings for test-center live exams, 97% of undercover test-takers carried printed notes or other prohibited materials through check-in, and 100% of those who brought in notes or other prohibited printed materials accessed those notes during the exam without being stopped. In remote testing live exams, 91% kept printed notes within reach and used them unnoticed, and 31% left camera view for an unauthorized break and returned without challenge.
Caveon’s findings reinforce the need for modernized, system-level approaches to exam integrity, such as its Observer platform, which incorporates advanced AI and machine learning technology with real-time analytics to surface meaningful risk indicators and trigger targeted human review only when predefined thresholds are exceeded. When paired with Caveon’s Scorpion test development and delivery platform, which offers secure exam design tools, such as the dynamic SmartItem® and randomly parallel tests, the Observer model addresses both visible and invisible forms of test fraud that traditional proctoring cannot detect.
“Proctoring was never designed to carry the full weight of exam security,” says David Foster, founder and CEO of Caveon. “When programs rely on observation alone, they create blind spots that undermine validity. Protecting exam outcomes requires systems that generate evidence, scale reliably and enhance human judgment rather than replace it.”
