What happened
The UK government appointed Jonathan Fisher KC to review disclosure in criminal cases and fraud offenses. Part 1 of his review, signed in November 2024, was published on 20 March 2025. It describes OpenText Axcelerate, which the Serious Fraud Office uses to find material that might undermine the prosecution or assist the defense. He writes that the tool finds such material at much faster rates than a human. Once trained, it uses machine learning to put the material most likely to meet the disclosure test at the top. A human reviewer then considers each item in that order. In February 2024 the SFO found that a configuration fault had left some documents out of search results. The tool was reconfigured and material re-analyzed, and he was told that nothing new of importance had been flagged so far. He concludes that such technology needs regular maintenance and rigorous performance evaluation. The review gives no figure for the speed.
Sources
-
Inquiry report, by authors with no tie to the system.
Jonathan Fisher KC, Independent Review of Disclosure and Fraud Offences. Disclosure in the Digital Age: Independent Review of Disclosure and Fraud Offences, Part 1 (CP 1285), published 20 March 2025, paras 231-233.
One such tool, used by the Serious Fraud Office (SFO) to identify undermine or assist material at much faster rates compared to a human, is OpenText Axcelerate.
-
Audit report, by authors with no tie to the system.
HM Crown Prosecution Service Inspectorate (HMCPSI). An inspection of the handling and management of disclosure in the Serious Fraud Office. Chapter 9, para 9.42 (also para 4.16 for the AI features)
There is no question that the move to Axcelerate brings an increased level of functionality and many more tools to better control and manage the disclosure process. In some interviews we heard from staff who were clearly using the increased levels of functionality to improve the efficiency and control of the disclosure process.
-
Audit report, by authors with no tie to the system.
HM Crown Prosecution Service Inspectorate (HMCPSI). Serious Fraud Office: case progression follow-up inspection. Para 10.27 (also para 7.33)
In the four of the five cases we examined, there was evidence that the Axcelerate system was being used. Since the 2019 inspection we can see that the new system is more effective for the handling and management of case material.
Details
AI system
- Workflow
-
Officers tag a portion of the files as relevant or not to train the tool. The tool ranks the material most likely to meet the disclosure test at the top. A human reviewer considers each item in the ranked order.
Follow-up
- Conditions for replication
-
The tool is configured correctly and checked often. Officers make accurate decisions early, since the tool learns from them.
Entry details
- Sector
-
Criminal justice / disclosure in fraud cases.
- Purpose
-
OpenText built Axcelerate to analyze digital material and rank documents for review, with machine learning trained on reviewers' decisions.
Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Axcelerate speeds the Serious Fraud Office's disclosure review (AIBD-0842). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0842.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)