AI TRAINER / DATA ANNOTATOR AGREEMENT

PREAMBLE

This AI Trainer / Data Annotator Agreement dated [Date] between [Company Name] ("AI Company") and [Trainer/Annotator] ("Trainer") governs provision of training data, annotations, and feedback for machine learning models. Legal Framework: GDPR 2016/679 (Data Protection), German BGB §611a (Services), US Copyright 17 USC (Fair Use), EU Copyright Directive 2019/790.

1.1 Training Data: Any data, images, text, audio, video, or other materials provided by Trainer for AI model development per GDPR Art. 4 (Definitions)
1.2 Annotations: Labels, classifications, corrections, feedback, or evaluations applied to training data
1.3 Personal Data: Any data identifying natural persons included in training data; processed per GDPR compliance framework
1.4 AI Company owns output models: Final trained models, weights, parameters belong exclusively to AI Company; Trainer retains no rights to model outputs

2.1 Authority: Trainer represents having authority to license training data per BGB §177 (Agency); data not subject to third-party IP rights/restrictions
2.2 Original Work: Training data is Trainer's original creation OR Trainer has obtained written permission from copyright/data owners per 17 USC §201 (Copyright Ownership)
2.3 No Personal Data Violations: Training data does NOT contain personal data of third parties without lawful processing basis per GDPR Art. 6 (Lawful Basis)

3.1 Training Data IP Assignment: Trainer grants AI Company: (a) Permanent, worldwide, royalty-free license to train models, (b) Right to create derivative works, (c) Right to commercialize models per UrhG §31 (License Grant)
3.2 Trainer IP Retention: Trainer retains copyright in original training data but cannot use data for competing AI purposes (non-compete per BGB §90)
3.3 No Attribution Required: AI Company not required to credit Trainer in published models or commercial products

4.1 Personal Data Processing: If training data contains personal data: (a) Processing lawful basis = [consent / legitimate interest / performance of contract] per GDPR Art. 6, (b) Trainer ensures data subject rights compliance, (c) AI Company acts as processor per GDPR Art. 28 (Data Processing Agreement)
4.2 Data Subject Rights: Trainer warrants personal data subjects have been notified of AI training use per GDPR Art. 13-14 (Transparency)
4.3 Data Retention: AI Company retains training data for [X years] for model improvement; deletion upon termination (unless archival required)

5.1 Prohibited Training Data: Trainer shall NOT provide:

• ❌ Copyrighted content without license (books, movies, music - exceptions per 17 USC §107 (Fair Use))
• ❌ Personal data without consent/lawful basis (GDPR Art. 6 violations)
• ❌ Trade secrets of third parties per EU Trade Secrets Directive
• ❌ Biometric data of non-consenting individuals per GDPR Art. 9 (Special Categories)
• ❌ Content used to discriminate/harass per EU Race/Ethnicity Directive

5.2 Fair Use Defense: AI Company's use of training data protected under fair use doctrine (transformative use, research purposes) per 17 USC §107

6.1 Payment Structure: AI Company pays Trainer:

• ☐ Fixed fee per annotation: EUR/USD [X per item]
• ☐ Per-hour rate: EUR/USD [X per hour] (tracked via [time-tracking system])
• ☐ Flat project fee: EUR/USD [X total] for [deliverables]

6.2 Payment Terms: Payment due [Net 30] days after invoice; late fees [1.5% monthly] per German Late Payment Act
6.3 No Royalties: Trainer waives right to royalties from model commercial use; one-time payment constitutes full compensation

7.1 Confidentiality Duty: Trainer maintains strict confidentiality regarding: (a) Model architecture/weights, (b) Training methodology, (c) Commercial strategy per EU Trade Secrets Directive
7.2 Duration: Confidentiality obligation survives engagement termination for [5 years] or trade secret protection period (indefinite if truly secret)
7.3 Exception: Trainer may disclose to government agencies (legal disclosure) or competent courts

8.1 Trainer Indemnity: Trainer indemnifies AI Company from claims:

• Copyright/patent infringement in training data per 17 USC
• Privacy violations from personal data per GDPR Art. 82 (Damages)
• Trade secret misappropriation per Trade Secrets Directive Art. 4

8.2 Liability Cap: Trainer liability capped at total fees paid (12 months max) per BGB §280

9.1 Termination for Cause: AI Company may immediately terminate if Trainer: (a) breaches confidentiality, (b) provides infringing data, (c) violates GDPR obligations
9.2 Post-Termination: Upon termination: (a) Trainer ceases all work immediately, (b) All materials returned/destroyed, (c) Final payment within 30 days if no breach
9.3 Survival: Sections 3 (IP Ownership), 4 (Data Protection), 7 (Confidentiality), 8 (Indemnification) survive termination indefinitely

Law: ☐ German (GDPR + BGB) ☐ [US State] | Disputes: Mediation → Binding arbitration per DIS Rules