AI Disclosure

1. What Top5 does

Top5 is an AI-assisted CV screening service. When a recruiter uploads a CV, Top5:

• extracts text from the CV and computes a vector embedding (OpenAI)
• ranks candidates against a job description using vector similarity
• for selected candidates, runs a Topgrading-style structured assessment (Anthropic Claude)
• presents the recruiter with scores, recommendations, and a shortlist they can review and override

Top5 does not auto-reject candidates, auto-hire candidates, or take any decision on its own. Every score, recommendation, or shortlist Top5 produces is advisory and is subject to human review by the recruiter who chose to use Top5.

2. How AI is used, and by whom

Top5 uses AI supplied by third-party model providers. We do not build or train our own models. There are two distinct uses, and each sees a different slice of data:

• Semantic search. A third-party embedding provider (OpenAI) converts CV text into a numeric representation so CVs can be ranked against a role description by meaning rather than keyword. It produces no scores and no recommendations.
• Topgrading assessment. A third-party language-model provider (Anthropic) receives CV text and the job description and returns the competency scores, evidence and recommendation the recruiter then reviews.

Both providers are named, with their processing region, on our sub-processor list.

We do not publish the exact model identifiers or versions we use. Internally we pin the model version in code and record it with every assessment, and when we change the model or the prompt we increment the assessment version so the audit trail can distinguish runs. That record is available to regulators and to customers under our AI Act technical file; it is not published on this page, because a public version string goes stale the moment a provider deprecates it and is not information a candidate can act on.

We do not train AI models on your CV data or candidate data. CVs are sent to the model providers above only to generate the immediate response, under contractual "no training" terms with each provider.

3. Limitations and risks

AI assessment is imperfect. You should expect:

• scores that are sensitive to CV formatting, length, and writing style
• weaker signal for candidates with non-traditional career paths, career gaps, or sparse CVs
• residual bias from the training data of the underlying models — we cannot fully eliminate this
• language-bias: assessment quality is highest for English-language CVs
• occasional model errors (hallucinated facts, wrong scores) that human review must catch

4. EU AI Act — your rights (Article 13 transparency)

For EU candidates, Top5 falls within Annex III(4)(a) of the EU AI Act (Regulation 2024/1689) as a high-risk AI system used in employment. As a candidate processed by Top5 you have the right to:

• be informed when AI is being used to assess your CV
• obtain meaningful information about the logic involved in the assessment
• request human review of any AI-generated outcome that affects you (Article 14)
• contest an AI-generated outcome and seek explanation of the result
• be free from solely automated decisions with legal or similarly significant effects (GDPR Article 22)

To exercise these rights, contact the employer or recruiter who uploaded your CV in the first instance. They are the data controller under GDPR and the deployer under the EU AI Act. You may also contact us directly at privacy@top5cv.com and we will route your request.

5. UAE PDPL — automated processing rights

UAE candidates are protected by UAE Federal Decree-Law No. 45 of 2021 (PDPL). Where automated processing produces legal or significant effects, you have the right to:

• request human intervention in the decision
• express your point of view
• contest the decision and request an explanation

Contact privacy@top5cv.com or dpo@top5cv.com to exercise these rights.

6. NYC Local Law 144 — restricted use

Top5 does not currently permit unaudited use for hiring covered by NYC Local Law 144. See our NYC AEDT Notice for the full position.

If you are an NYC-resident candidate and you believe Top5 has been used to screen your application for an NYC-covered role, contact privacy@top5cv.com. We will investigate and contact the customer responsible.

7. Provider obligations (for customers)

LEFTFIELD SOLUTIONS FZE is the EU AI Act provider of Top5 and meets the following provider obligations under the self-assessment route in Article 43:

• Internal Annex IV technical documentation (available to enterprise customers on request)
• Risk-management and quality-management system, proportionate to the scale of Top5
• Post-market monitoring via our api_usage log and incident handling process
• Versioned prompt and model pins; assessment version is recorded on every candidate row
• Human-oversight requirement contractually imposed on customers via our Terms of Service
• Cooperation with relevant supervisory authorities on request

Customer-facing deployer guidance: see legal@top5cv.com to request our Customer AI-Use Guide.

8. Sub-processors

The list of third-party processors that Top5 relies on is published at /legal/sub-processors. Each sub-processor is bound by a data processing agreement or equivalent commitment.

9. Contact

Privacy and data subject rights: privacy@top5cv.com
Data Protection contact: dpo@top5cv.com
Legal and DPA requests: legal@top5cv.com