Politics · United Kingdom · 1 hr ago
Home Office asylum AI summaries fail 22% of initial accuracy checks
A rights group says Home Office AI summaries often contain errors in the UK’s asylum system.
The tools help officials summarise asylum interviews and search country policy information, and both use OpenAI’s GPT-4o model through Microsoft’s Azure service.
Experts rejected 45 of 203 summaries checked after the summarising tool began operating in April 2026, a 22% failure rate.
The group says checks covered only an estimated 0.3% of summaries from April to July, while testing of the separate policy search tool had stopped by August.
Applicants are not specifically told when AI is used and cannot inspect or correct the summaries of their interviews.
The group also found no clear evidence of faster decisions: the average monthly number of initial decisions fell about 42% compared with the same period in 2025, while average interview-to-decision time rose 29%, though this does not prove AI caused the changes.
The group wants both tools suspended and independently audited because it says current safeguards are inadequate.
A report by the Open Rights Group says 45 of 203 AI-generated asylum interview summaries failed initial expert accuracy checks after the system was rolled out in April 2026.
The 22 percent failure rate was higher than the 9 percent rejection rate recorded during the original pilot.
The group says checks covered only 0.3 percent of summaries from April to July and is calling for both Home Office AI tools to be suspended and independently audited.
The Home Office says humans remain responsible for asylum decisions, but the group questions whether caseworkers can reliably identify errors.
The report also found that asylum processing times worsened after the tools were introduced, though it does not prove the AI caused the slowdown.
- Who
- The Home Office and the Open Rights Group (ORG).
- What
- ORG reported that 22 percent of the AI-generated asylum interview summaries in its initial checks failed accuracy tests. It called for the tools to be suspended and independently audited.
- When
- The report was published on 13 October 2026. The figures cover the rollout from April 2026.
- Where
- The United Kingdom.
- Why
- The report raised concerns about errors, limited oversight and whether the tools are delivering promised efficiency gains.
Open Rights Group
Home Office
Accuracy
Open Rights Group
ORG says 45 of 203 summaries failed initial expert checks and warns that summaries can misrepresent what applicants said.
Home Office
The Home Office says humans remain responsible for asylum decisions.
Oversight
Open Rights Group
ORG says safeguards are inadequate and questions whether caseworkers can reliably spot mistakes.
Home Office
The Home Office's stated position is that humans remain responsible for decisions.
Use of the tools
Open Rights Group
ORG calls for both tools to be suspended and independently audited.
Home Office
The Home Office is using the tools to support caseworkers; it did not respond to The Register's questions.
The Home Office rolled out the Asylum Case Summarisation system.
The Home Office confirmed that expert testing of the Asylum Policy Search tool had stopped.
The Register published the report on the tools' accuracy and oversight.
- Initial failures
- 45 of 203 summaries
- Initial failure rate
- 22 percent
- Original pilot rejection rate
- 9 percent
- Further review
- 19 summaries flagged
- Second-stage reviews
- 30 of 56 failed the required standard
- Estimated share checked
- 0.3 percent from April to July









