Skip to product information
1 of 3

Svenska Dokumentmallar

AI Social Scoring 2026 – Prohibited Goods Screening, Customer Scoring, Supplier Certificates and Stop Decisions | Word/PDF/Excel

AI Social Scoring 2026 – Prohibited Goods Screening, Customer Scoring, Supplier Certificates and Stop Decisions | Word/PDF/Excel

File format
DOCX, PDF, XLSX
Document language
5 Swedish work templates and Swedish user guide; separate English User Guide (instructions only); Swedish Excel workbook.
Number of files
7

Digital download. No physical product is shipped.

Regular price 79 SEK
Regular price Sale price 79 SEK
Sale Sold out
Taxes included.
View full details

About this template

Does your company use AI to score, rank, or classify people? Check whether an actual use case risks falling under the prohibition of unlawful social scoring in Article 5.1 c of EU AI Regulation (EU) 2024/1689. This specialist package structures data sources, customer scores, negative consequences, supplier responses, and technical stops with traceable decisions.

Quick facts: EU law and Swedish business · prohibition applicable since 2 February 2025 · 5 Swedish work templates, Swedish user guide, and separate English User Guide · 7 editable Word (DOCX) files, 7 corresponding searchable PDFs, and 1 Excel (XLSX) file with 9 work tabs. 15 real customer files in a digital ZIP file after manual delivery activation. PDFs are not interactive forms.

01 · Package contents

Five standalone Swedish work templates for prohibition assessment, data sources, supplier certificates, consequences, and stop/correction decisions; a Swedish guide and a separate English guide. All seven documents are included in DOCX and matching PDF formats. A separate Swedish XLSX with nine tabs is also included: 7 Word + 7 PDF + 1 Excel = 15 files.

02 · What the files contain

Assessment of AI systems and the assessment of natural persons over time; how social behaviour and personal traits are used; comparison between the original and new context; negative treatment, relevance, and proportionality; verifiable supplier information; separate technical stop verification, and a proper human decision. Excel provides internal signals, never an automatic legal approval.

Who is this package for?

Example 1 – AI for customer decisions: A digital platform uses an AI-generated behaviour score to deny a person a service that has no connection to the context in which the data was collected. In this case, a concrete Article 5.1 c assessment is needed, not just a general AI policy.

Example 2 – purchasing a scoring service: A store or service provider needs to check what an external AI model actually scores, what personal traits it infers, and whether the effect could be unjustified or disproportionate. The supplier letter and consequence template provide a basis for an actual assessment.

Important: A standard bonus or loyalty programme, a risk assessment using relevant transaction data, or a lawful credit assessment is not automatically prohibited social scoring. Even if Article 5.1 c does not apply to a system, GDPR, consumer law, non-discrimination law, or specific AI Act rules may still apply.

Exactly what is included in the ZIP file?

File Purpose Language Format
01 · Prohibition assessment AI systems, assessment over time, social behaviour/personal traits, and two alternative branches of harm SV DOCX/PDF
02 · Data sources and context Collection context, reuse, profiling, and evidence of data sources SV DOCX/PDF
03 · Supplier letter and certificates Request actual evidence of the model, data source, and configuration from the supplier SV DOCX/PDF
04 · Consequence assessment Customer treatment, negative outcomes, proportionality, and legitimate comparative cases SV DOCX/PDF
05 · Stop and correction decisions Technical block, retesting, verified action, and separate management attestation SV DOCX/PDF
06 · Swedish user guide Workflow, legal boundaries, and concrete scenarios SV DOCX/PDF
07 · English User Guide English instructions for Swedish forms, not English legal templates EN DOCX/PDF
08 · Excel control register Nine tabs for systems, data sources, scoring models, requirements, and actions SV XLSX

The main templates and Excel are in Swedish. The English User Guide is a separate guide on how the Swedish documents are used, not an English version of the work templates.

How Excel works – nine tabs

The Excel register contains Overview, System, Data Sources, Scoring Models, Consequences, Requirements Assessment, Suppliers, Actions, and Read Me. The workbook is delivered without filled-in cases; the eight key metrics show 0. It has input options, formulas, and control signals that check whether all basic requirements and at least one of the two alternative branches of harm have been marked with supporting evidence.

Example: When an actual AI system assesses people over time based on social behaviour and the resulting score can lead to negative treatment in an unrelated context, POSSIBLE PROHIBITION – STOP / ASSESS is displayed. If a basic requirement is missing, ART 5 C NOT ESTABLISHED – OTHER LAW is displayed. The latter is not a green light for deployment. An action marked as “Completed” does not become “verified” without separate test evidence and a verification mark.

14 scenario tests have been carried out. Excel does not analyse real AI models or personal profiles, does not make decisions about legality, and does not replace a GDPR impact assessment. Use pseudonymous internal evidence and case IDs; do not enter real sensitive personal data into a shared Excel file.

How to use the templates

  1. Identify the actual AI system, version, supplier, and your role as supplier or user.
  2. Document whether people/groups are assessed over time based on social behaviour or personal/personality traits.
  3. Map the original context of the data sources and which decisions the score controls.
  4. Assess whether negative treatment occurs in an unrelated social context or is unjustified/disproportionate.
  5. Request testable supplier evidence, especially if the score comes from an external model or platform.
  6. In case of suspected prohibited use: document the technical stop, actual effect, necessary correction, and separate verification.
  7. Let the appropriate person in charge assess whether to continue. Human approval or customer consent does not override a genuine prohibition.

Legal: what exactly does Article 5.1 c say?

The cumulative basis of the prohibition: AI systems placed on the market, put into service, or used for the assessment or classification of natural persons/groups over time based on social behaviour or known, inferred, or predicted personal or personality traits, when the resulting social score leads to unfavourable treatment of any of these types:

  • Option (i): unfavourable treatment in a social context that has no connection to the context in which the data originally arose or was collected.
  • Option (ii): unfavourable treatment that is unjustified or disproportionate in relation to the social behaviour or its severity.

The basic requirements must be assessed together, while it is sufficient if at least one of the two harm options occurs. The prohibition covers both public and private actors and has been applicable since 2 February 2025. EU Regulation 2024/1689, Article 5.1 c and Recital 31.

Boundaries: The European Commission's guidelines distinguish between prohibited social scoring and legitimate, purpose-specific assessment that complies with EU and national law, such as certain forms of creditworthiness assessment, fraud risk assessment, and insurance underwriting. Such use may also be subject to other requirements. The guidelines are guidance, not a binding court interpretation. European Commission guidelines C(2025)5052, Section 4.

Compare with the store's nearby templates

For general AI use and internal AI rules, there is the AI policy for companies. For AI-based indiscriminate collection of facial images, there is the specific template AI Act Article 5.1 e – Facial Databases. For the human right to review in automated credit assessment, there is Automated credit assessment. This product instead covers the assessment of unlawful social scoring under Article 5.1 c.

Frequently asked questions

Are AI-based customer scores prohibited?

No, not generally. The actual basic requirements and at least one of the two alternative negative outcomes must be assessed. Standard scoring programmes or legitimate purpose-specific scoring are not automatically covered.

Can companies use a social score if the customer has consented?

Consent cannot in itself make a use covered by a substantive prohibition in the AI Regulation permissible. The legal basis for processing personal data must also be assessed separately.

Is a credit score the same as social scoring?

Not necessarily. Relevant economic data for a legitimate credit assessment may fall outside this prohibition, but credit law, GDPR, and the rules for high-risk AI may still be relevant.

Can Excel determine whether my AI system is lawful?

No. The formulas provide conservative internal signals and require a responsible person to assess real data, consequences, legal boundaries, and a potential stop.

What do we do if the supplier does not provide sufficient supporting evidence?

Record the missing evidence, escalate responsibility, and consider pausing the use in question until the actual risks have been assessed. The supplier's marketing claims are not sufficient as a reliable conclusion.

Are all the work documents also in English?

No. The five work templates and Excel are in Swedish. A Swedish guide and a separate English User Guide are included. The English instructions are not full translations of the main templates.

Are the PDF files interactively fillable?

No. Use DOCX for editing and PDF for searchable reading/printing.

Sources, version, and legal quality

Version 1.0 · AI-assisted source review 15 November 2026. Official sources: AI Regulation (EU) 2024/1689, European Commission Article 5, Recital 31, European Commission non-binding interpretive guidelines, and GDPR. No external legal review, regulatory certification, or guarantee of legal compliance has been performed. For a concrete implementation, legal conclusions and other relevant rules must be reviewed separately.