Article 6(3) Non-High-Risk Determination Memo

BrightPath HR Solutions GmbH ("the Provider"), a human resources technology company established and operating within the European Union, develops and deploys the AI system designated "ResumeSort Assistant," version 1.3, released in April 2026. The System is designed and intended to serve as an AI-powered document organiser that automatically categorises and tags incoming job applications by advertised role, department, and stated qualifications. It pre-sorts applications into structured review folders for subsequent examination by human recruiters employed by the Provider's enterprise clients across eight EU Member States.

The System does not rank, score, evaluate, or filter candidates, and produces no output that constitutes or influences an employment-related decision concerning any natural person. The operational context of the System is confined to the preparatory administrative phase of the recruitment workflow.

The intended users of the System are HR recruitment teams operating within the Provider's enterprise client organisations. These users interact with the System's output — the categorised folders and associated metadata tags — as a workflow efficiency tool. The System is not deployed in a consumer-facing capacity and does not interact directly with job applicants.

This document constitutes a formal Article 6(3) Non-High-Risk Determination Memo prepared in accordance with Regulation (EU) 2024/1689. Its purpose is to record the Provider's reasoned assessment that the System satisfies the conditions set out in Article 6(3) and is therefore not classified as a high-risk AI system. The memo is maintained as an internal governance record and is available for inspection by any competent national supervisory authority upon request.

Annex III to Regulation (EU) 2024/1689, Paragraph 4(a), specifically captures "AI systems intended to be used for the recruitment or selection of natural persons, in particular to place targeted job advertisements, to analyse and filter job applications, and to evaluate candidates." The legislative intent underpinning this category is the protection of natural persons from AI-driven decisions that may adversely affect their access to employment.

The System operates within the recruitment process: it receives job applications submitted by natural persons and performs automated processing upon those documents. The fact that the System's processing is limited to organisational categorisation rather than evaluative assessment does not, at the prima facie stage, remove it from the scope of Annex III, paragraph 4(a). The Regulation's language — "to analyse and filter job applications" — is broad enough on its face to encompass a system that parses, categorises, and routes application documents.

The Provider therefore finds, for the purposes of this determination, that the System falls within the prima facie scope of Annex III, paragraph 4(a). This triggers the obligation to assess whether the conditions for the Article 6(3) exception are satisfied.

Article 6(3) provides that an AI system falling within Annex III shall not be considered high-risk where it performs one of four specified functions and does not pose a significant risk of harm to the health, safety, or fundamental rights of natural persons.

The System falls within the prima facie scope of Annex III, paragraph 4(a) by virtue of its deployment within the recruitment workflow and its automated processing of job application documents. However, the System satisfies the conditions for the Article 6(3) exception, principally under criterion (d) and additionally under criterion (a), and does not pose a significant risk of harm to the health, safety, or fundamental rights of natural persons.

The System's function is strictly preparatory and organisational: it categorises incoming job applications by role, department, and stated qualifications, and routes them into structured review folders. It does not evaluate, rank, score, compare, recommend, or filter candidates. The design of the System deliberately excludes any evaluative capability, and the Provider has imposed architectural constraints that prevent the System from producing scoring or ranking outputs.

Human recruiters employed by the Provider's enterprise clients retain exclusive authority over all substantive candidate assessment. The System's organisational output does not constrain, bias, or influence the exercise of that authority.

Notwithstanding the non-high-risk determination, the Provider acknowledges and accepts the residual obligations imposed by Regulation (EU) 2024/1689 on all providers of AI systems, regardless of risk classification.

Article 50 transparency obligations: The System does not interact directly with job applicants; it operates as a back-office document processing tool accessed only by the Provider's enterprise client recruitment teams. Accordingly, Article 50(1) is not directly engaged in respect of applicant-facing interactions. The Provider commits to ensuring that enterprise clients are informed of the System's role in the application-handling process. The Article 50 transparency compliance date of 2 August 2026 applies to all applicable obligations and has been incorporated into the Provider's compliance programme.

GPAI obligations: The System does not incorporate or rely upon a general-purpose AI model within the meaning of Chapter V. The categorisation and tagging functionality is implemented through purpose-built classification algorithms. Chapter V obligations are accordingly not engaged.

The Provider adopts proportionate governance measures described in Section 6 of this memo, in keeping with its commitment to responsible AI deployment and in anticipation of the compliance deadlines established by the Regulation, including the Annex III high-risk compliance date of 2 December 2027.

Responsible Officer: To be designated by BrightPath HR Solutions GmbH management.

Review Schedule: This memo must be reviewed and reaffirmed on an annual basis, and immediately upon any material change to the System's design, training data, output capabilities, or intended purpose. The non-high-risk determination is conditional upon the System continuing to satisfy the criteria assessed in Section 3 above. Any modification that introduces evaluative, ranking, or scoring functionality would require re-assessment and may render the Article 6(3) exception inapplicable.

Triggers for Re-assessment:

• — Addition of candidate ranking, scoring, or comparative assessment features
• — Change in the System's output from administrative routing to evaluative recommendation
• — Reduction of human oversight over candidate evaluation
• — Supervisory authority guidance materially affecting the Article 6(3) analysis
• — Amendment to Annex III or to the interpretation of Article 6(3) criteria

Documentation Retention: This memo must be retained for a minimum of 10 years from the date of determination, alongside any supporting technical documentation and governance records.

Version: 1.0 — April 2026