Fundamental Rights Impact Assessment (FRIA): AI Act Guide

Who needs a FRIA under Article 27 of the EU AI Act, what it must contain, how it fits with a DPIA, and a practical template.
On this page

Last updated: 16 November 2026

Quick answer: A fundamental rights impact assessment (FRIA) is the assessment Article 27 of the EU AI Act requires from public bodies, private providers of public services, and lenders and life or health insurers before they first use certain high-risk AI systems. It documents use, affected people, risks, human oversight and remedies, and applies from 2 December 2027.

Most data protection officers already know how to run a DPIA. The EU AI Act adds a second, related assessment for a narrower group of organisations: the fundamental rights impact assessment. It looks beyond personal data to the full range of rights an AI system can affect, from non-discrimination to access to public services.

The Digital Omnibus on AI moved the high-risk timetable in 2026, which gives deployers extra time. This guide explains who needs a FRIA, what it must contain, how it fits with a GDPR DPIA, and a practical template you can adapt now.

What is a fundamental rights impact assessment under the EU AI Act?

A fundamental rights impact assessment is a structured, documented review that certain deployers must complete before putting a high-risk AI system into use. Its purpose is to identify how the system could harm people’s fundamental rights in your specific context and what you will do about it.

The obligation sits in Article 27 of Regulation (EU) 2024/1689, the AI Act. Providers assess risks when they build a system; the FRIA is the deployer’s assessment of real-world use.

Who must carry out a FRIA?

Three groups of deployers must perform a FRIA before deploying a high-risk AI system listed in Annex III: bodies governed by public law, private entities providing public services, and deployers using AI for credit scoring or for life and health insurance risk assessment and pricing.

Article 27(1) sets out the scope precisely. It applies to high-risk systems referred to in Article 6(2), which are the stand-alone use cases in Annex III, with one exception: systems intended for use in critical infrastructure (point 2 of Annex III) are excluded.

Deployer typeFRIA required?Typical examples
Body governed by public lawYes, for any Annex III system except critical infrastructureMinistries, municipalities, public universities, public hospitals
Private entity providing public servicesYes, for any Annex III system except critical infrastructureOutsourced benefits administration, private operators of public education or health services
Any deployer using creditworthiness or credit scoring AI (Annex III, 5(b))YesBanks, consumer lenders, buy-now-pay-later providers
Any deployer using AI for life and health insurance risk assessment and pricing (Annex III, 5(c))YesLife and health insurers
Other private deployers of Annex III systemsNo FRIA; Article 26 duties applyA retailer screening job applicants

Annex III 5(b) expressly excludes AI systems used to detect financial fraud, so a fraud detection model does not trigger a FRIA under that point.

Common mistake: Assuming the FRIA only concerns the public sector. A private bank using an AI credit scoring model for consumer loans is in scope, even if it has no public functions at all.

When does the FRIA obligation apply?

The FRIA obligation applies from 2 December 2027, the new date for stand-alone Annex III high-risk systems set by the Digital Omnibus on AI. The assessment must be completed before first use, so preparation needs to start well before that date.

The Digital Omnibus on AI, Regulation (EU) 2026/1744, entered into force on 27 July 2026. The European Commission’s AI Act overview now lists 2 December 2027 for high-risk systems in sensitive areas such as employment, education and credit, and 2 August 2028 for high-risk AI embedded in regulated products.

Under Article 27(2), the FRIA applies to the first use of the system. You may rely on a previously conducted FRIA, or on an existing impact assessment carried out by the provider, in similar cases. If any element changes or is no longer up to date during use, you must update the information.

What must a FRIA contain?

Article 27(1) lists six mandatory elements. Every FRIA must describe the deployment process, the period and frequency of use, the people affected, the specific risks of harm, the human oversight measures, and the measures to take if risks materialise.

  1. Deployment process (27(1)(a)): a description of the deployer’s processes in which the system will be used, in line with its intended purpose.
  2. Period and frequency (27(1)(b)): how long and how often the system is intended to be used.
  3. Affected persons (27(1)(c)): the categories of natural persons and groups likely to be affected in the specific context.
  4. Specific risks of harm (27(1)(d)): the risks likely to affect those categories, taking into account the information the provider gives under Article 13.
  5. Human oversight (27(1)(e)): how oversight measures will be implemented according to the instructions for use.
  6. Mitigation and remedies (27(1)(f)): the measures to take if those risks materialise, including internal governance arrangements and complaint mechanisms.

Point (d) is where most of the analytical work sits. The provider’s instructions for use should describe known limitations, performance for specific groups and foreseeable misuse. Your job is to translate that into risk for your own population.

Does a FRIA replace a GDPR DPIA?

No. A FRIA and a DPIA are separate obligations, but Article 27(4) says that where a DPIA under Article 35 GDPR already meets any of the FRIA obligations, the FRIA complements it. You do not need to repeat the same analysis twice.

Article 26(9) of the AI Act also requires deployers to use the provider’s information to carry out a DPIA where GDPR requires one. The two work best as a single workflow with shared inputs. Our guide to DPIAs under GDPR covers the data protection side in detail.

AspectGDPR DPIA (Article 35)AI Act FRIA (Article 27)
TriggerProcessing likely to result in high risk to individualsFirst use of an Annex III high-risk system by an in-scope deployer
WhoControllerDeployer (public bodies, public service providers, lenders, life and health insurers)
Rights coveredMainly privacy and data protectionAll fundamental rights, including non-discrimination and effective remedy
Regulator contactPrior consultation with the supervisory authority if high residual riskNotify the market surveillance authority of the results using the template
UpdateReview when the risk changesUpdate when any element changes or is out of date

Running both from the same record of the system saves time. Enactia’s AI-powered GRC platform links DPIA workflows to your record of processing activities and AI Act requirements, so one system description feeds both assessments. To try it on one of your AI use cases, start your 14-day free trial.

How do you notify the market surveillance authority?

Once the FRIA is complete, Article 27(3) requires the deployer to notify the market surveillance authority of its results, submitting the filled-out template as part of the notification. Deployers may be exempt from notifying in the exceptional cases covered by Article 46(1).

Article 27(5) tasks the AI Office with developing a template questionnaire, including through an automated tool, to help deployers comply in a simplified way. Build your internal template so it can be mapped onto the official questionnaire.

FRIA template: a practical structure

Use the structure below as a working FRIA template. It covers the six Article 27 elements and adds the governance fields a DPO or AI governance lead will need for accountability.

SectionWhat to recordSource of information
1. System identificationSystem name, provider, version, Annex III category, ownerAI inventory, contract
2. Deployment processBusiness process, decisions supported, level of automationProcess owner, workflow maps
3. Period and frequencyStart date, duration, volume of decisions per periodProject plan
4. Affected personsCategories of people and groups, including vulnerable groupsCustomer or citizen data, DPIA
5. Rights at riskEach relevant Charter right and how it could be affectedLegal and DPO analysis
6. Risks of harmLikelihood and severity per group, known limitationsProvider instructions for use, testing
7. Human oversightNamed overseers, competence, authority to overrideArticle 26(2) records
8. Mitigation and remediesControls, escalation, complaint route, suspension criteriaRisk register, complaints procedure
9. LinksDPIA reference, vendor assessment, previous FRIAsDocument repository
10. Sign-off and notificationApprover, date, notification reference, review dateGovernance records

Worked example: a consumer credit scoring model

Consider a mid-sized bank replacing its scorecard with an AI credit scoring model in 2027. The deployment process is consumer loan applications through online and branch channels, with the model recommending approve, refer or decline. Affected persons include all applicants, with particular attention to young applicants, recent migrants and self-employed people whose credit histories are short or irregular.

The main risks are indirect discrimination, opaque refusals that people cannot challenge, and over-reliance on the model by staff. Mitigations could include a mandatory human review of every decline, a documented override process, adverse action explanations, periodic outcome testing by group, and a clear complaint route. The DPIA already covers lawful basis, data minimisation and security, so the FRIA cross-references it rather than repeating it.

In practice: Ask the provider during procurement for accuracy and error data by relevant subgroup. Without it, you cannot complete point (d) with any confidence, and your vendor risk assessment should treat its absence as a finding.

How should a DPO prepare now?

Start with the systems you already use or plan to buy, confirm which are Annex III high-risk, and check whether your organisation falls into one of the three FRIA groups. Then set up a template and workflow that connects with your existing DPIA process.

  • Build or update your AI inventory and tag every Annex III use case.
  • Confirm whether you are a body governed by public law, a private provider of public services, or a user of 5(b) or 5(c) systems.
  • Add FRIA requirements to AI procurement and contract templates.
  • Align the FRIA template with your DPIA and enterprise risk register.

An AI management system helps keep this repeatable. Our ISO 42001 certification roadmap shows how an AI management system can host impact assessments alongside other AI governance processes.

Key takeaways

  • A FRIA is required under Article 27 before first use of an Annex III high-risk AI system by public bodies, private providers of public services, and credit scoring or life and health insurance deployers.
  • Critical infrastructure systems are excluded, and fraud detection is outside Annex III point 5(b).
  • After the Digital Omnibus, the obligation applies from 2 December 2027.
  • A FRIA has six mandatory elements and its results must be notified to the market surveillance authority.
  • It complements, but does not replace, a GDPR DPIA, so run both from shared inputs.

Frequently asked questions

What is a fundamental rights impact assessment?
It is an assessment required by Article 27 of the EU AI Act, completed by certain deployers before they first use a high-risk AI system listed in Annex III. It describes how the system will be used, who it affects, the specific risks to their fundamental rights, human oversight measures, and the governance and complaint measures to apply if risks materialise.

Do private companies need to carry out a FRIA?
Some do. Private entities providing public services must carry out a FRIA for Annex III systems other than critical infrastructure. Any deployer using AI to assess creditworthiness or set credit scores, or for risk assessment and pricing in life and health insurance, must also do one. Other private deployers of high-risk systems follow Article 26 instead.

When does the FRIA requirement start?
Following the Digital Omnibus on AI, Regulation (EU) 2026/1744, which entered into force on 27 July 2026, the high-risk obligations for Annex III systems, including Article 27, apply from 2 December 2027. The FRIA must be done before first use, so organisations planning deployments around that date should complete it during procurement and testing.

Can a DPIA be reused for a FRIA?
Partly. Article 27(4) states that where a DPIA under Article 35 GDPR already meets any of the FRIA obligations, the FRIA complements it. You can cross-reference shared content, such as the system description and data flows, but you still need to address rights beyond data protection, such as non-discrimination and access to remedies.

Who receives the FRIA results?
The deployer notifies the market surveillance authority of the results, submitting the filled-out template developed by the AI Office as part of the notification. In the exceptional situations covered by Article 46(1), deployers may be exempt from this notification. Keep the full assessment on file so you can show it to authorities on request.

Make the FRIA part of your AI governance routine

The fundamental rights impact assessment will be new to many organisations, but the building blocks are familiar: a clear description of processing, an honest risk analysis, named owners and a remedy for people who are affected. DPOs who connect the FRIA to their existing DPIA, vendor and risk processes now will be ready well before December 2027.

To see how one platform can run FRIAs, DPIAs and AI Act controls together, contact us or book a demo or start your 14-day free trial.

This article is for general information and is not legal advice.

Written by
Share
Built for MSPs

Turn compliance into recurring revenue. Launch in days, not months.

Governance, risk & compliance made simple

Fewer spreadsheets. Faster compliance. See Enactia in action.

Ready to make compliance simple?

See how Enactia brings governance, risk and compliance into one AI-powered platform. Book a demo now.