In March 2026, the Association of Chartered Certified Accountants (ACCA) stopped offering remote sittings for its main qualification, sending most of its 500,000 students back to in-person exam centers. ACCA's chief executive, Helen Brand, told The Financial Times that the shift is a response to rising misconduct, noting that "people who want to do bad things are probably working at a quicker pace" as AI tools make remote supervision harder to trust.
For certification bodies, universities, and corporate training programs running high-stakes exams outside a supervised room, the real question is no longer whether remote assessment is convenient. It is whether the proctoring behind it can be trusted when it matters most.
What breaks down under real exam-day pressure
A single camera feed cannot catch a second device, a mirrored screen, or a coached earpiece
Exam formats vary. Oral defenses, paper-based sittings, and closed-book multiple choice do not all fit the same proctoring mode
Sitting windows concentrate thousands of candidates into the same few hours, and volume spikes are exactly when weak systems fail
What does AI proctoring software actually do?
AI proctoring software monitors a remote exam session for signs that the registered candidate is not the person taking the test, or that they are getting help they should not have. It combines identity checks, behavior monitoring, and device detection, then routes anything unusual to a human reviewer rather than deciding guilt on its own. The goal is not full automation. It is giving a small proctoring team the reach to cover exam volumes that would otherwise require a room full of supervisors.
Why generic video monitoring is not built for this
Facial recognition alone misses a lot. Device detection that covers phones and headphones, not just cameras, screen-mirroring detection, and virtual machine checks close the gaps that a single detection method leaves open. A secure browser that locks down the exam environment itself adds a layer no amount of webcam monitoring can replace.
Modality breadth matters just as much as detection depth. Oral exams, paper-based exams, and no-camera formats all need proctoring too, and forcing every format through one webcam-only mode leaves gaps exactly where candidates know to look for them.
What to check before choosing a vendor
Any vendor can put a big number on a slide. Before signing, ask for the details that separate a proven platform from a projected one.
- A concurrent-session figure the vendor has actually delivered, not projected
- Independently held certifications such as ISO 27001, GDPR, and FERPA, not self-reported claims
- Support for the exam formats you actually run, including oral and paper-based sittings, not just webcam
- Device detection that covers phones and headphones, since cameras alone will not catch either
How Constructor Proctor is built for this
Constructor Proctor supports more than 10,000 simultaneous exam sessions today, live and AI-assisted modes on the same platform, and device detection that covers both phones and headphones.
Its secure browser catches screen mirroring and unauthorized devices without needing constant human review. Unified live proctoring covers paper-based and oral exams alongside standard webcam sessions, so certification bodies are not forced onto a different vendor for every exam format.
10,000+
Simultaneous exam sessions supported
90%
Accuracy in detecting violations
ISO 27001
GDPR and FERPA certified
That reliability matters at scale. TechNet has used Constructor Proctor to deliver more than 80,000 professional certification exams, with capacity for over 1,000 exams a day and a 70% reduction in exam deliberation time. The Irish Aviation Authority also uses Constructor Proctor and Assess for theoretical remote pilot assessments under EU aviation regulations, showing how the same infrastructure can support regulated assessment environments with strict integrity requirements.
What our customers say





Frequently asked questions
AI proctoring software monitors remote exam sessions for identity mismatches, unauthorized devices, and unusual behavior, then flags anything suspicious for a human reviewer instead of deciding automatically.
A single human proctor can realistically watch a small number of candidates at once. AI-assisted monitoring applies the same detection criteria consistently across thousands of sessions at the same time, which matters most during concentrated exam windows.
No. Responsible platforms use AI to handle continuous monitoring at scale, then route flagged incidents to trained human reviewers who apply context before any consequences are decided.
Live and AI-assisted webcam sessions, oral exams, and paper-based sittings, all on the same platform, so certification bodies do not need a separate vendor for each format.
Discover Constructor Proctor
See how Constructor Proctor keeps a human in the loop and the audit trail an ICO review expects.
