The UK exams regulator has imposed a £875,000 penalty on Cambridge English after automated marking errors in the International English Language Testing System (IELTS) led to incorrect results being issued to more than 62,000 test takers over a two-year period.
The findings, published by Ofqual, reveal that systemic weaknesses in automated scoring processes allowed errors to go undetected between August 2023 and September 2025, affecting one of the world’s most widely used English language qualifications.
Automated marking failures across millions of tests
IELTS, jointly owned by British Council, IDP Education, and Cambridge University Press & Assessment under the wider governance of University of Cambridge, is taken globally for university admissions, professional registration, and visa applications.
According to Ofqual’s penalty notice, automated scoring errors occurred in the listening and reading components of the test, which are computer-marked based on predefined answer sets. The issue affected specific “gap match” question types, where candidates select words from a list to complete sentences.
Over the affected period, nearly 7.8 million IELTS test instances were processed. Of these, 93,865 responses were incorrectly marked due to technical faults, including issues where correct answers were marked wrong and incorrect answers marked correct.
Impact on candidates and results
Of the affected cases, 62,794 learners received incorrect component or overall results that were later corrected. Around 20,602 candidates saw their scores increase after re-marking, while 1,115 experienced a downward adjustment.
Most score changes were minor—typically a 0.5 adjustment on the IELTS 0–9 band scale—though a small number saw a full-band increase.
Importantly, 1,108 affected candidates had taken the Secure English Language Test (SELT), the version required for UK visa and immigration purposes. Of these, four cases were found to have impacted visa eligibility decisions. All four individuals subsequently retook the test and met the required requirements.
Ofqual stated there was no evidence of wider systemic harm beyond the directly affected cases, though it acknowledged limitations in available data.
Systemic weaknesses and delayed detection
Regulators attributed the failure to “systemic weaknesses” in Cambridge English’s automated marking oversight, including inadequate monitoring and error-detection systems. The issue remained undetected for an extended period and was only identified during an internal system update.
The faults included misordered answer keys during data transfer between systems and incorrect handling of diacritical marks such as accents and umlauts, which should have been ignored but sometimes led to incorrect scoring outcomes.
Financial penalty and remediation
While imposing the fine, Ofqual noted Cambridge English’s extensive remediation efforts, including more than £6 million spent on refunds, compensation, customer support, and system upgrades.
In total, 26,246 test takers requested and received refunds. Cambridge English also established a 24/7 support hub and implemented new safeguards intended to prevent similar failures in the future.
A spokesperson for IELTS said the organisation accepted responsibility, apologised to affected candidates, and acted quickly once the issue was identified, offering refunds or resits to all impacted test takers.
Implications for international students and institutions
The incident highlights growing scrutiny of large-scale digital assessment systems used in high-stakes qualifications, particularly those linked to university admissions and immigration pathways.
IELTS remains one of the most widely accepted English proficiency tests worldwide, but the case raises questions about reliance on automated scoring systems in contexts where results can directly affect education and visa outcomes.
As global mobility continues to depend heavily on standardised language testing, regulators and institutions face increasing pressure to ensure that automation is both accurate and robust enough to protect candidates from systemic error.
- Log in to post comments