Skip to main content

FRIA template AI Act: what the Commission's draft contains and lacks

FRIA template AI Act: what the Commission's draft of 16 November 2026 contains, what is still blank, and how the amended Art 27(4) DPIA route works.

By Scanara

Key Takeaways

  1. 1.A fundamental rights impact assessment (FRIA) checks, before deployment, how an AI system may affect fundamental rights. Art 27(1) names two groups of deployers (those "using an AI system under its authority", Art 3(4)) that must perform one. Group one is "bodies governed by public law" or "private entities providing public services". For the first, the explanatory notice says the concept "may be understood as covering State organs and regional and local authorities". Group two is deployers of the systems in "points 5 (b) and (c) of Annex III". These cover natural persons' credit scores and life and health insurance. Annex III is the Act's list of high-risk areas of use.
  2. 2.As of 19 November 2026 the draft names no way to submit a notification, and three links in it read "[link to be added later]". Its cover says it "has not been adopted or endorsed by the European Commission".
  3. 3.The draft says it "follows the structure of Article 27(1) AI Act". It adds Low, Medium and High ticks for the likelihood and severity of each risk.
  4. 4.Art 27(4) says a deployer "may" include cross-references to its data protection impact assessment (DPIA), where an obligation is already met through it. Regulation (EU) 2026/1744, the Digital Omnibus on AI, which amends the AI Act, replaced the paragraph. A DPIA is a data protection impact assessment under EU data protection law.

On 16 November 2026 the European Commission published a draft explanatory notice and template (PDF) for the FRIA under the EU AI Act. It announced the draft in a Commission news item on the draft FRIA template. "The draft" means the Commission's draft explanatory notice and template. "The final template" means the template under Art 27(5).

The draft is for deployers. A deployer is any "natural or legal person, public authority, agency or other body using an AI system under its authority" (Art 3(4)). Art 3(4) excludes systems used "in the course of a personal non-professional activity".

A deployer in scope must perform a FRIA before it deploys a high-risk AI system referred to in Article 6(2) (Art 27(1)). Art 99(1) has Member States lay down "the rules on penalties and other enforcement measures" "applicable to any infringement of this Regulation by operators". The list in Art 99(4) of provisions subject to administrative fines names deployer obligations "pursuant to Article 26". Art 27 is not named in that list. Which penalties or other enforcement measures apply to a missing FRIA is for each Member State to set under Art 99(1).

Art 27(1) excludes Annex III point 2, critical infrastructure. Under the AI Act as consolidated on 27 July 2026, Art 27 applies to Annex III systems from 2 December 2027 (Art 113, point (c)(i)).

The deployer must notify the market surveillance authority of the result (Art 27(3)). The market surveillance authority is "the national authority carrying out the activities and taking the measures pursuant to Regulation (EU) 2019/1020" (Art 3(26)). See Which authority supervises the AI Act depends on the system and on where you sell.

As of 19 November 2026, the news item of 16 November 2026 invites deployers to express interest in testing the draft to the AI Office by 4 December 2026. Art 3(47) defines the AI Office. It is "the Commission's function of contributing to the implementation, monitoring and supervision of AI systems and general-purpose AI models, and AI governance". Testing is voluntary.

Next step: read the draft, find your group, and compare one of your systems with its seven sections. As of 19 November 2026, the news item of 16 November 2026 sets 15 January 2027 as the end of the feedback period.

Art 27(1) names two groups of deployers and excludes Annex III point 2

Group two has no public or private qualifier: any deployer of these systems is covered. Art 27(1) reads:

Prior to deploying a high-risk AI system referred to in Article 6(2), with the exception of high-risk AI systems intended to be used in the area listed in point 2 of Annex III, deployers that are bodies governed by public law, or are private entities providing public services, and deployers of high-risk AI systems referred to in points 5 (b) and (c) of Annex III, shall perform an assessment of the impact on fundamental rights that the use of such system may produce.

— Art 27(1)

Group two is deployers of the systems in "points 5 (b) and (c) of Annex III". Point 5 (b) covers the creditworthiness or credit score of natural persons. Point 5 (c) covers risk assessment and pricing in relation to natural persons for life and health insurance. Point 5 (b) carries this exception: "with the exception of AI systems used for the purpose of detecting financial fraud".

The draft's explanatory notice (pages 2 to 5 of the draft) repeats both groups and the point 2 exclusion. It gives its own reading of group one, "without prejudice to the interpretation of the Court of Justice". It states: "The AI Act does not define the concept of 'bodies governed by public law'." It says the concept "may be understood as covering State organs and regional and local authorities, including legislative, executive, administrative, law enforcement and judicial authorities". It adds that the concept may also cover other entities established under public law to perform public tasks. It adds that the concept may also cover entities entrusted by law with the exercise of public authority, and Union institutions, bodies, offices and agencies.

For private entities, the explanatory notice quotes Recital 96 of the Act: public services are "services important for individuals that are of public nature". Its examples are education, healthcare, social services, housing and the administration of justice. It says these examples suggest that private entities providing other services of a similar public nature should conduct a FRIA.

Art 27(3) lets deployers be exempt from the notification in the Art 46(1) case

Art 27(3) says: "In the case referred to in Article 46(1), deployers may be exempt from that obligation to notify." Art 46(1) lets a market surveillance authority, upon a duly justified request, "authorise the placing on the market or the putting into service of specific high-risk AI systems within the territory of the Member State concerned, for exceptional reasons of public security or the protection of life and health of persons, environmental protection or the protection of key industrial and infrastructural assets". The explanatory notice states: "This exemption concerns the notification only, not the obligation to conduct the FRIA itself."

The draft follows Art 27(1)(a) to (f) in seven sections and adds its own risk scale

Sections 2 to 6 of the draft follow Art 27(1)(a) to (f) in order, and the explanatory notice says "This template follows the structure of Article 27(1) AI Act".

  • Section 1, General information, asks for the organisation, its type and a responsible person. It asks for two EU database numbers: the AI system's identification number and the deployer's registration number (if applicable). It also asks for the version and date. A tick list covers the areas of use, "(excluding point 2)". It asks whether this is the first deployment and whether you rely on a previous FRIA or the provider's assessment (Art 27(2)). It also asks whether this updates an earlier FRIA and which stakeholders you consulted.
  • Section 2, Context of use (Art 27(1)(a) and (b)), asks for the intended purpose, the processes or tasks, and the period and frequency of use.
  • Section 3, Affected categories (Art 27(1)(c)), is a tick list of eleven entries, the last being "Other (please specify)". Among them are children, older people, persons with disabilities and third country nationals, and "The general public (does not replace the groups above)".
  • Section 4, Fundamental rights impact and risks (Art 27(1)(d)), has two parts. Table 4.1, headed "Risk assessment related to the right to the protection of personal data", covers that right and, to the extent a DPIA assessed them, other fundamental rights. Block 4.2, headed "Assessment of risks of harm to other fundamental rights", takes one block per risk. Each block has a tick list of rights, which also includes the right to the protection of personal data. It also asks for the affected groups, likelihood, severity, mitigation measures "(not required for Low risk)" and the residual level.
  • Section 5, Human oversight (Art 27(1)(e)), asks for the measures "in accordance with the instructions for use", who oversees, the authority to override and the tools. It also asks about competence, training, authority and support (Art 26(2)) and AI literacy (Art 4).
  • Section 6, Measures if risks materialise (Art 27(1)(f)), asks for monitoring, reporting and response, including suspension and reporting to the provider under Art 26(5). It also asks for complaints handling, internal governance and redress.
  • Section 7, Overall conclusion, asks for a short synthesis of the risks from Section 4. A Submission block then asks for a date of submission, a signature, and a name and function (Art 27(3)).
  • A final summary box is for deployers that register their use of the AI system in the EU database (Art 49(3)). These are public authorities, Union institutions, bodies, offices or agencies, and persons acting on their behalf.

The risk scale is the draft's own addition. Likelihood and severity each take one of three ticks, Low, Medium or High. The draft defines them: "Low: unlikely under foreseeable conditions. Medium: plausible under some conditions. High: highly probable or already observed in comparable deployments." The draft has no numeric score, no weighting and no pass or fail. Its level-dependent rules are that measures are "not required for Low risk" in block 4.2 and "not where both likelihood and severity are Low" in table 4.1.

The draft names no submission route, leaves further guidance for later and has no automated tool yet

As of 19 November 2026 the draft names no way to submit a notification, because its notification link reads "[link to be added later]".

  • Status: the cover says "This draft has not been adopted or endorsed by the European Commission."
  • Three links read "[link to be added later]". They are the update link ("This can be done here [link to be added later]."), the advice link ("please consider reaching out to competent authorities that can provide advice, such as the [link to be added later]") and the notification link ("Deployers can do so by [link to be added later].").
  • Guidance: the explanatory notice says guidance and background information "may in the future be provided in info boxes next to specific questions or in accompanying guidance". Sections 5.1 and 5.2 of the draft already carry short "Guidance:" paragraphs.
  • Section 7: the draft says a function to generate the synthesis automatically will be considered for the automated tool.
  • Automated tool: the news item of 16 November 2026 says "Once adopted, the template will also be set up as an automated tool on the AI Act Single Information Platform".
  • Authorities: the draft does not name the competent market surveillance authority of each Member State.

How a DPIA and a FRIA relate: the amended Art 27(4) lets a deployer cross-refer to a DPIA

Art 27(4) says a deployer may include cross-references to a DPIA in its FRIA, where an obligation is already met through it. Regulation (EU) 2026/1744 replaced paragraphs 4 and 5 of Art 27 and changed no other paragraph. In the consolidated version of 27 July 2026, paragraph 4 reads:

If any of the obligations laid down in this Article is already met through the data protection impact assessment conducted pursuant to Article 35 of Regulation (EU) 2016/679 or Article 27 of Directive (EU) 2016/680, the deployer may, when conducting the fundamental rights impact assessment referred to in paragraph 1 of this Article, include cross-references to the relevant sections of that data protection impact assessment or include relevant parts thereof in the fundamental rights impact assessment.

— Art 27(4)

The wording is a permission, and it applies to an obligation only if that obligation "is already met through" the DPIA. Texts written before Regulation (EU) 2026/1744 may quote the earlier paragraph 4, which no longer applies.

Art 27(5) says of the final template: "This template shall, where relevant, give deployers the possibility to include cross-references to the relevant sections of the data protection impact assessment or include relevant parts thereof in the fundamental rights impact assessment pursuant to paragraph 4."

The explanatory notice states: "The FRIA complements the data protection impact assessment (DPIA) conducted under EU data protection law." It then states: "The DPIA looks at risks raised by the personal data processing activities to the rights and freedoms of data subjects." Art 27(1) frames the FRIA around "the impact on fundamental rights that the use of such system may produce".

Table 4.1 of the draft asks whether a DPIA exists. It names three laws. The first is the General Data Protection Regulation (GDPR, Article 35). The second is the Law Enforcement Directive (Directive (EU) 2016/680, Article 27). The third is the EU Data Protection Regulation for Union institutions (EUDPR, Article 39). It says "you may cross-refer to the DPIA (and attach relevant section(s)) or incorporate its relevant parts".

Six steps a deployer can take before Art 27 applies

Art 27 applies to Annex III systems from 2 December 2027 (Art 113, point (c)(i)), so a deployer has time to read, check, map, reuse, test and comment.

  1. Read the draft explanatory notice and template (PDF) and the Commission news item of 16 November 2026. The explanatory notice covers who must do a FRIA, when to conduct it and how to document it.
  2. Check whether you are in group one or group two (Art 27(1)). If you are a private entity, compare your service with the explanatory notice's examples. Take advice from counsel if the answer is unclear, because the notice's reading is "without prejudice to the interpretation of the Court of Justice".
  3. Map one system against the draft's seven sections. Collect the provider's instructions for use, because Art 27(1)(d) refers to "the information given by the provider pursuant to Article 13" and Art 27(1)(e) to "the instructions for use". The explanatory notice lists the provider's instructions for use among its preparation steps: "obtained, reviewed and ... kept available".
  4. Reuse an existing DPIA where it already meets an Art 27 obligation, and cross-refer to it (Art 27(4)).
  5. As of 19 November 2026, optional: express interest in testing the draft by 4 December 2026 through the EU Survey testing form. Whether the form still accepts answers was not tested. The news item says the AI Office is looking for deployers that will have to carry out a FRIA under Art 27. Participants apply the draft to one of their own systems, and the AI Office supports them with an introductory session, replies to questions and a closing session. The AI Office "may have to select the cases for the testing, ensuring a balanced representation of areas of use, use cases, types and sizes of deployers, and Member States".
  6. As of 19 November 2026, the news item of 16 November 2026 asks for comments by 15 January 2027 through the EU Survey feedback form on the draft.

Scanara structures your Art 27(1) answers and exports a notification document, but it is not the Commission's draft

In the Scanara product (state of 19 November 2026), the FRIA assessment is a guided questionnaire. Its questions cover the six elements of Art 27(1) and also ask about reviews, incident escalation and your DPIA.

Scanara drafts a FRIA report and a notification document and prefills sections from your answers. The notification dossier must be created and generated first.

In the web app you can start a new version of an approved FRIA. It starts from the current questions with your earlier answers carried over, and the earlier version stays as it was.

Scanara exports the notification document for an approved FRIA as a download (PDF by default). It sends nothing to an authority; you send it yourself.

Scanara's questions and documents are its own structure, not the Commission's draft and not the template referred to in Art 27(3) and (5).

Scanara does not test the Art 27(1) scope for you. You decide whether you are in group one or group two.

Related Reading

Sources & References

Frequently Asked Questions

Is the Commission's draft FRIA template binding?

No. As of 19 November 2026 the PDF says "This draft has not been adopted or endorsed by the European Commission." Art 27(5) tells the AI Office to "develop a template for a questionnaire".

Which deployers must do a FRIA?

Art 27(1) names deployers that "are bodies governed by public law, or are private entities providing public services". It also names deployers of systems in "points 5 (b) and (c) of Annex III". Art 27(1) excludes Annex III point 2.

Does a DPIA replace a FRIA?

Art 27(4) does not say so. It says a deployer "may" include cross-references to the relevant sections of the DPIA, or include relevant parts of it, in the FRIA.

When must a deployer notify the authority?

Art 27(3) says "Once the assessment referred to in paragraph 1 of this Article has been performed, the deployer shall notify the market surveillance authority of its results". Art 27 applies to Annex III systems from 2 December 2027 (Art 113, point (c)(i)). As of 19 November 2026 the draft names no submission route.

Must a deployer test the draft by 4 December 2026?

No. The news item of 16 November 2026 says "Participation is voluntary and without prejudice to obligations under the AI Act."

Scanara prepares evidence from your repositories

Scanara reads your connected GitHub repositories and runs rule-based scans for standard AI and machine-learning patterns. It can miss non-standard code. Separately, it generates draft documentation from templates and your assessment answers as evidence for your compliance work. You make the classification decisions.