IRCC Spousal Sponsorship Advanced Analytics: A 2026 Transparency Dataset and Comparative Review
A source-by-source comparison of what IRCC publicly discloses about advanced analytics in in-Canada and overseas spouse/partner processing, what those disclosures do not establish, and where human decision-making remains.
Executive finding. IRCC publicly discloses automated positive eligibility processing for routine spouse/partner cases in both the in-Canada and overseas systems reviewed here. Admissibility and the final decision remain with an officer, and IRCC says the disclosed tools do not refuse or recommend refusing applications.
What IRCC says the tools do
For routine cases within the disclosed systems, IRCC says positive eligibility can be approved through automation for the sponsor and principal applicant. Cases not positively approved through automation proceed to officer review.
The overseas disclosure is more detailed in several areas. IRCC describes separate tools for sponsor and principal-applicant review and says those tools use officer-developed rules and machine-learning-generated rules based on prior IRCC files. The overseas notice also identifies safeguards including review for potential discriminatory impacts, Gender-Based Analysis Plus, a quality-assurance plan, privacy and security elements, regular monitoring and officer override.
Human review and final decisions
The public disclosures preserve a human-decision boundary. Positive eligibility may be automated for routine cases, but admissibility and the final decision remain with an officer. For overseas spouse/partner processing, IRCC states that the tools never refuse or recommend refusing applications and that applications not positively approved through the tools are routed to manual officer review.
A material limitation in the inland record
The Office of the Auditor General reported that cases not streamlined by the tool were facing longer processing times. IRCC said it was monitoring differential processing outcomes between cases processed through automation and cases decided by officers. IRCC also stated that manual review should not be treated as an adverse signal.
Inland and overseas comparison
| Publicly disclosed field | In-Canada spouse/common-law | Overseas spouse/partner |
|---|---|---|
| Current IRCC classification | Automated decision making | Automated decision making |
| Routine positive eligibility automation | Yes | Yes |
| Sponsor scope | Yes | Yes |
| Principal-applicant scope | Yes | Yes |
| Refusal or refusal recommendation by disclosed tool | IRCC says no | IRCC says no |
| Route when positive automation does not apply | Full individualized officer assessment | Manual officer review |
| Admissibility | Officer | Officer |
| Final decision | Officer | Officer |
| Publicly described rules basis | Officer-developed rules plus advanced analytics | Officer-developed rules plus machine-learning-generated rules based on prior IRCC files |
| Separate triage function in current IRCC inventory | Not separately listed | Yes |
| Public impact wording | Level 2; uses AI | Moderate |
| GBA+ | Not publicly established in the reviewed inland source set | Publicly described |
| Quality-assurance plan | Not publicly established in the reviewed inland source set | Publicly described |
| Privacy/security-by-design detail | Not publicly established in the reviewed inland source set | Publicly described |
| Officer override | Publicly described | Publicly described |
| Peer-review resource | Not listed in the inland Open Government record reviewed | Official resource listed; findings not summarized in v1.0 |
“Not publicly established” does not mean “No.” It means the reviewed public source set does not establish that proposition.
What the public records do not establish
- hidden model features or variables;
- relationship-evidence scoring weights;
- undisclosed fraud or risk scores;
- officer preferences or secret refusal criteria;
- why a particular file was or was not automated;
- whether a PFL is generated by a disclosed analytics tool;
- an individual applicant’s approval or refusal probability;
- whether a particular type or quantity of relationship evidence increases the chance of automated approval; or
- whether an individual applicant will receive faster or slower processing because of the disclosed analytics.
Applicants and representatives should not treat public transparency records as instructions for gaming a system. The records describe parts of IRCC’s processing environment; they do not replace complete, coherent and truthful application evidence or case-specific professional advice where appropriate.
Methodology
This release uses official Canada.ca and Open Government sources for propositions about IRCC systems and processing. Each dataset field is tied to an official source ledger. In-Canada and overseas systems are treated separately, and historical disclosure dates are separated from current 2026 operational context. No confidential client records, case outcomes or unpublished MAK case data are used.
Official source set
- IRCC technology inventory
- In-Canada Open Government record
- Overseas Open Government record
- IRCC overseas advanced-analytics notice
- IRCC/PACP differential-processing response
- IRCC Q-1772 AIA material
- IRCC 2026 Family Reunification transition material
Peer-review disclosure
The overseas Open Government record lists a public Peer Review Summary - Overseas Spouse and Partner Advanced Analytics Project (English resource ID ad8dd581-6a9c-40f1-ae56-1ed2f0698701). Its existence and resource metadata are verified. The detailed report text has not been deterministically extracted and checked in this release, so v1.0 does not summarize or attribute findings to that peer review.
Professional-review boundary
Professional review: Omer Khalil, RCIC R710149. Omer professionally reviewed the v0.2 interpretation from which this v1.0 release is derived. This is professional RCIC review, not academic peer review and not Government of Canada endorsement.
Machine-readable release data
The complete v1.0 dataset is now deposited publicly on Zenodo in CSV and JSON formats. The version-specific persistent identifier is DOI 10.5281/zenodo.22542743.
The public CSV and JSON use the same 28 record fields. Evidence-bearing values preserve distinctions such as NOT_PUBLICLY_ESTABLISHED; that state does not mean “No.” The Zenodo DOI provides persistent citation identity. It is not academic peer review, Government endorsement or independent editorial endorsement.
Reuse and citation
Copyright in MAK-created original expression, comparison structure, normalization, methodology, commentary and presentation is retained by MAK Canadian Immigration Services. Ordinary citation, linking and short quotation are permitted subject to applicable law and accurate attribution. This release is not placed under a blanket Creative Commons/open licence and does not grant a blanket commercial-republication, bulk-redistribution or model-training licence for the complete MAK-created package. For selected MAK-created tables, rows, CSV/JSON records and derived analysis, supplemental reuse permissions are published in MAK Research Reuse and Citation Permissions v1.0.
Suggested citation: MAK Canadian Immigration Services. “IRCC Spousal Sponsorship Advanced Analytics: A 2026 Transparency Dataset and Comparative Review.” Version 1.0, September 6, 2026. DOI: 10.5281/zenodo.22542743.
Government of Canada and other third-party source materials remain subject to their own licences and terms.
Version note: v1.0 is the first authorized public release. It is derived from the validated v0.2 internal source object, incorporates the September 6, 2026 source recheck and clarified Office of the Auditor General attribution, and preserves the validated non-claims and inland/overseas separation.
