evidoria

← Back to browse

Good practice Imported

PIGR — Portugal's Social Security Launches an AI Platform to Flag Fraud Risk Before It Happens

Portugal · Lisbon · See the Portugal profile · See the Lisbon profile

Evidence: Descriptive / self-reported Top 93% 27/100 · Ask Evidence Copilot about this practice

Portugal's Social Security went live in October 2025 with an AI platform that scans internal records for behavioural risk patterns among employers, aiming to catch fraud before it happens. A citizen-facing module is planned; no results are public yet.

200,000 companies
Companies affected by the related digital contribution-cycle automation (from 1 January 2026)
shortened from 12 to 3 months
Presumption period for detecting fraudulently backdated employment records (effective 1 January 2026)
PIGR — Portugal's Social Security Launches an AI Platform to Flag Fraud Risk Before It Happens

Details

Maturity
Pilot
Promoter
Instituto de Informática, I.P. / Segurança Social
Period
2025-ongoing
Region (NUTS)
PT17
Keywords
social security, fraud prevention, public finance

Context

In October 2025, Portugal's Segurança Social put into operation the Plataforma Integrada de Gestão do Risco (PIGR), built in-house by the Instituto de Informática, I.P. The system applies AI/machine learning to data already held in Social Security's own information systems to detect behavioural risk patterns among employer entities.

Objectives

PIGR aims to flag suspicious cases for early intervention rather than only investigating fraud after the fact.

Activities

The first module, covering employer entities, is live; a second module extending to individual citizens is planned but not yet launched. A related administrative reform effective 1 January 2026 will automate the digital contribution cycle for around 200,000 companies and shorten the presumption period used to catch fraudulently backdated employment records from twelve months to three.

Conclusions

Neither Social Security nor Portuguese press coverage report any performance statistics for PIGR: no figures on cases flagged, false-positive rates, or fraud value prevented have been made public. The EU AI Act classifies this kind of eligibility- and fraud-related public-sector AI system as 'high-risk', requiring an impact assessment and regular audits; no such published assessment was found. This is an early-stage, evidentially opaque deployment worth tracking rather than a proven success.

Implementation
Indicative cost
Low (< €50k)
Time to results
Short (< 1 year)
Staffing & skills
Instituto de Informática, I.P. (in-house development team), Segurança Social operational staff handling flagged cases

Conditions for success

  • Publication of performance statistics (cases flagged, false-positive rates, fraud value prevented)
  • Completion and publication of the EU AI Act-mandated high-risk impact assessment and regular audits
  • Successful, transparent launch of the planned second module covering individual citizens

Common failure modes

  • No performance statistics published by Social Security or Portuguese press coverage
  • No published EU AI Act high-risk impact assessment despite the system's fraud/eligibility classification
  • Second module (individual citizens) not yet launched

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful