AI DATA PROCESSING AGREEMENT
AGREEMENT DATE: [DATE]
AGREEMENT NUMBER: [DPA-NUMBER]
PARTIES
DATA CONTROLLER ("Controller"):
- Legal Name: [CONTROLLER LEGAL NAME]
- Address: [FULL ADDRESS]
- Contact: [NAME, EMAIL, PHONE]
DATA PROCESSOR ("Processor"):
- Legal Name: [PROCESSOR LEGAL NAME]
- Address: [FULL ADDRESS]
- Contact: [NAME, EMAIL, PHONE]
- Data Protection Contact: [DPO NAME, EMAIL]
RECITALS
WHEREAS, Controller and Processor have entered into an agreement for AI-related services ("Principal Agreement");
WHEREAS, Processor will process Personal Data on behalf of Controller using AI systems in connection with the Principal Agreement;
WHEREAS, the parties wish to ensure that the processing of Personal Data by AI systems complies with applicable data protection laws including GDPR, CCPA/CPRA, and other privacy regulations;
WHEREAS, AI systems present unique data processing considerations including training data usage, model inputs and outputs, and algorithmic decision-making;
NOW, THEREFORE, the parties agree as follows:
ARTICLE 1: DEFINITIONS
1.1 "AI System" means the artificial intelligence system(s) used by Processor to provide services under the Principal Agreement, as described in Schedule A.
1.2 "Applicable Data Protection Laws" means all laws and regulations relating to the processing of Personal Data, including GDPR, CCPA/CPRA, and other applicable privacy laws.
1.3 "Data Subject" means an identified or identifiable natural person whose Personal Data is processed.
1.4 "Input Data" means Personal Data provided to the AI System as inputs (prompts, queries, data for analysis).
1.5 "Model Training" means the process of using data to develop, train, fine-tune, or improve AI models.
1.6 "Output Data" means data generated by the AI System that may contain or be derived from Personal Data.
1.7 "Personal Data" means any information relating to an identified or identifiable natural person, as defined under Applicable Data Protection Laws.
1.8 "Processing" means any operation performed on Personal Data, including collection, use, storage, disclosure, and deletion.
1.9 "Security Incident" means any unauthorized access, acquisition, use, or disclosure of Personal Data.
1.10 "Special Category Data" means sensitive personal data as defined under GDPR Article 9, including data revealing racial or ethnic origin, political opinions, religious beliefs, health data, biometric data, and data concerning sex life or sexual orientation.
1.11 "Sub-processor" means any third party engaged by Processor to process Personal Data on behalf of Controller.
ARTICLE 2: SCOPE OF PROCESSING
2.1 Subject Matter and Duration
This DPA governs Personal Data processing in connection with:
- Subject Matter: [DESCRIBE AI SERVICES]
- Duration: Duration of the Principal Agreement
- Nature of Processing: [DESCRIBE PROCESSING ACTIVITIES]
2.2 Categories of Data Subjects
| Category | Description |
|---|---|
| [CATEGORY 1] | [DESCRIPTION] |
| [CATEGORY 2] | [DESCRIPTION] |
| [CATEGORY 3] | [DESCRIPTION] |
2.3 Categories of Personal Data
| Category | Data Elements | Purpose |
|---|---|---|
| [CATEGORY 1] | [ELEMENTS] | [PURPOSE] |
| [CATEGORY 2] | [ELEMENTS] | [PURPOSE] |
| [CATEGORY 3] | [ELEMENTS] | [PURPOSE] |
2.4 Special Category Data
☐ Special Category Data will NOT be processed
☐ Special Category Data will be processed as follows:
| Type | Purpose | Safeguards |
|---|---|---|
| [TYPE] | [PURPOSE] | [SAFEGUARDS] |
2.5 AI-Specific Processing
Input Data Processing:
☐ Personal Data provided as AI System inputs
☐ Inputs processed for: [PURPOSE]
☐ Input retention: [RETENTION PERIOD]
Output Data Processing:
☐ Outputs may contain/reflect Personal Data
☐ Outputs used for: [PURPOSE]
☐ Output retention: [RETENTION PERIOD]
Model Training:
☐ Personal Data will NOT be used for Model Training
☐ Personal Data MAY be used for Model Training only with explicit consent/legal basis
☐ Only anonymized/aggregated data used for Model Training
ARTICLE 3: CONTROLLER OBLIGATIONS
3.1 Lawful Processing
Controller warrants that:
☐ Processing has a lawful basis under Applicable Data Protection Laws
☐ Data Subjects have been provided required information/notices
☐ Necessary consents have been obtained (if consent is the legal basis)
☐ Controller has authority to provide Personal Data to Processor
3.2 Instructions
Controller shall provide documented instructions for Processing, including:
☐ Purposes of Processing
☐ Types of Personal Data to be processed
☐ Duration of Processing
☐ Special requirements or restrictions
3.3 Data Subject Rights
Controller is responsible for:
☐ Responding to Data Subject rights requests
☐ Notifying Processor of relevant requests
☐ Providing information to Data Subjects about AI processing
3.4 AI-Specific Notices
Controller shall ensure Data Subjects are informed about:
☐ Use of AI in processing their data
☐ Automated decision-making (per GDPR Article 22)
☐ Logic involved in AI processing
☐ Significance and consequences of AI processing
ARTICLE 4: PROCESSOR OBLIGATIONS
4.1 Processing Limitations
Processor shall:
☐ Process Personal Data only on documented instructions from Controller
☐ Not process Personal Data for own purposes
☐ Not use Personal Data for Model Training without explicit authorization
☐ Promptly inform Controller if instructions appear unlawful
4.2 Confidentiality
Processor shall:
☐ Ensure personnel are bound by confidentiality obligations
☐ Limit access to authorized personnel only
☐ Train personnel on data protection requirements
4.3 Security Measures
Processor shall implement appropriate technical and organizational measures including:
☐ Encryption of Personal Data at rest and in transit
☐ Access controls and authentication
☐ Regular security testing
☐ Incident detection and response capabilities
☐ Business continuity measures
AI-Specific Security:
☐ Protection against adversarial attacks
☐ Prevention of model inversion/extraction attacks
☐ Secure handling of AI inputs and outputs
☐ Logging of AI processing activities
4.4 Sub-processors
(a) Controller's authorization for Sub-processors:
☐ General authorization with notification of changes
☐ Specific authorization required for each Sub-processor
(b) Current Sub-processors listed in Schedule B
(c) Processor shall:
- Impose data protection obligations on Sub-processors
- Remain liable for Sub-processor actions
- Notify Controller of Sub-processor changes with [DAYS] notice
4.5 International Transfers
Personal Data shall be transferred outside [EEA/JURISDICTION] only:
☐ To countries with adequacy decisions
☐ Subject to Standard Contractual Clauses (attached as Schedule C)
☐ Under other approved transfer mechanisms: [SPECIFY]
4.6 Data Subject Rights Assistance
Processor shall assist Controller with:
☐ Responding to Data Subject access requests
☐ Data portability requests
☐ Deletion/erasure requests
☐ Objections to automated decision-making
☐ Requests for human review of automated decisions
4.7 Security Incident Response
(a) Processor shall notify Controller of Security Incidents within [24/48/72] hours
(b) Notification shall include:
- Description of incident
- Categories and approximate number of Data Subjects affected
- Likely consequences
- Measures taken or proposed
(c) Processor shall cooperate with Controller's incident response
4.8 Data Protection Impact Assessment
Processor shall assist Controller with DPIAs by providing:
☐ Information about AI System processing
☐ Security measures implemented
☐ Risk assessment information
☐ Bias and fairness testing results
4.9 Compliance Demonstration
Processor shall:
☐ Make available information necessary for compliance demonstration
☐ Allow and contribute to audits and inspections
☐ Maintain records of processing activities
ARTICLE 5: AI-SPECIFIC PROVISIONS
5.1 AI Transparency
Processor shall provide Controller with:
☐ Description of AI System capabilities and limitations
☐ Information about AI model(s) used
☐ Explanation of how Personal Data is processed by AI
☐ Information needed for Data Subject transparency
5.2 Model Training Restrictions
Unless explicitly authorized in writing:
☐ Processor shall NOT use Controller's Personal Data for Model Training
☐ Processor shall NOT use Input Data for Model Training
☐ Processor shall NOT use Output Data for Model Training
☐ Processor shall NOT share Personal Data with third parties for training
If Model Training is authorized:
☐ Training limited to: [SPECIFY SCOPE]
☐ Anonymization/aggregation required: ☐ Yes ☐ No
☐ Opt-out mechanism available: ☐ Yes ☐ No
☐ Training data retention: [PERIOD]
5.3 Automated Decision-Making
If AI System makes automated decisions about Data Subjects:
☐ Processor provides meaningful information about logic involved
☐ Human review mechanism available
☐ Controller can request human intervention for specific decisions
☐ Processor supports Controller's GDPR Article 22 compliance
5.4 Input/Output Data Handling
Input Data:
☐ Input Data retained only as long as necessary for processing
☐ Input Data deleted: [TIMEFRAME]
☐ Input Data not logged/stored beyond immediate processing need
☐ Other: [SPECIFY]
Output Data:
☐ Output Data may contain Personal Data derivatives
☐ Output Data retention: [PERIOD]
☐ Controller responsible for Output Data once delivered
5.5 AI Bias and Fairness
Processor represents:
☐ AI System has been tested for bias
☐ Bias mitigation measures implemented
☐ Will notify Controller of discovered bias affecting Personal Data processing
☐ Will cooperate with Controller's fairness assessments
5.6 AI System Changes
Processor shall notify Controller before:
☐ Material changes to AI System processing Personal Data
☐ Changes affecting data protection measures
☐ Changes to AI model versions
☐ Changes to Sub-processors providing AI services
ARTICLE 6: DATA RETENTION AND DELETION
6.1 Retention Periods
| Data Type | Retention Period | Basis |
|---|---|---|
| Input Data | [PERIOD] | [BASIS] |
| Output Data | [PERIOD] | [BASIS] |
| Processing Logs | [PERIOD] | [BASIS] |
| Training Data (if applicable) | [PERIOD] | [BASIS] |
6.2 Deletion and Return
Upon termination or Controller request:
☐ Processor shall delete Personal Data within [DAYS]
☐ OR return Personal Data to Controller
☐ Certify deletion in writing
☐ Delete from AI training data (if applicable and technically feasible)
Exceptions: Processor may retain data required by law for [PERIOD].
6.3 Deletion from AI Systems
☐ Processor will delete Personal Data from AI System databases
☐ Personal Data in trained models: [HANDLING - e.g., not feasible to remove / will retrain without data]
☐ Logs containing Personal Data deleted per retention schedule
ARTICLE 7: AUDITS AND COMPLIANCE
7.1 Audit Rights
Controller may:
☐ Conduct audits of Processor's compliance
☐ Engage third-party auditors (subject to confidentiality)
☐ Review Processor's certifications and audit reports
☐ Request specific compliance information
7.2 Audit Procedures
☐ [DAYS] advance notice required
☐ Audits during normal business hours
☐ Scope limited to AI processing and data protection
☐ Controller bears audit costs (unless audit reveals material breach)
7.3 Certifications
Processor maintains:
| Certification | Scope | Expiration |
|---|---|---|
| [CERT 1] | [SCOPE] | [DATE] |
| [CERT 2] | [SCOPE] | [DATE] |
ARTICLE 8: LIABILITY AND INDEMNIFICATION
8.1 Liability Allocation
Each party is liable for damages caused by processing in violation of:
- This DPA
- Applicable Data Protection Laws
- Controller's lawful instructions
8.2 Indemnification
Processor shall indemnify Controller for:
- Processor's breach of this DPA
- Processor's breach of Applicable Data Protection Laws
- Sub-processor violations
Controller shall indemnify Processor for:
- Controller's breach of this DPA
- Unlawful instructions
- Controller's breach of Applicable Data Protection Laws
8.3 Limitation
Liability limitations in the Principal Agreement apply to this DPA, except:
☐ No limitation for willful or grossly negligent violations
☐ Regulatory fines not subject to limitation
ARTICLE 9: TERM AND TERMINATION
9.1 Term
This DPA commences on the Effective Date and continues for the duration of the Principal Agreement.
9.2 Survival
Confidentiality, data deletion, and indemnification provisions survive termination.
ARTICLE 10: GENERAL PROVISIONS
10.1 Governing Law
This DPA is governed by [JURISDICTION] law.
For EU data: GDPR provisions govern regardless of choice of law.
10.2 Conflict
In case of conflict between this DPA and the Principal Agreement regarding data protection, this DPA prevails.
10.3 Amendments
Amendments must be in writing signed by both parties.
10.4 Entire Agreement
This DPA and its Schedules constitute the entire data processing agreement.
SIGNATURES
CONTROLLER:
Signature: _________________________________
Name: [NAME]
Title: [TITLE]
Date: _________________________________
PROCESSOR:
Signature: _________________________________
Name: [NAME]
Title: [TITLE]
Date: _________________________________
SCHEDULE A: AI SYSTEM AND PROCESSING DESCRIPTION
A.1 AI System Description
| Field | Description |
|---|---|
| System Name | [NAME] |
| System Type | [TYPE] |
| Provider | [PROVIDER] |
| Purpose | [PURPOSE] |
A.2 Processing Activities
| Activity | Personal Data | Purpose | Retention |
|---|---|---|---|
| [ACTIVITY] | [DATA] | [PURPOSE] | [PERIOD] |
A.3 Technical Measures
[DESCRIBE TECHNICAL SECURITY MEASURES]
A.4 Organizational Measures
[DESCRIBE ORGANIZATIONAL SECURITY MEASURES]
SCHEDULE B: SUB-PROCESSORS
| Sub-processor | Location | Processing Activities | Safeguards |
|---|---|---|---|
| [NAME] | [LOCATION] | [ACTIVITIES] | [SAFEGUARDS] |
SCHEDULE C: STANDARD CONTRACTUAL CLAUSES
[ATTACH APPLICABLE SCCs FOR INTERNATIONAL TRANSFERS]
SCHEDULE D: TECHNICAL AND ORGANIZATIONAL MEASURES
D.1 Encryption
- At rest: [STANDARD]
- In transit: [STANDARD]
D.2 Access Control
- Authentication: [METHOD]
- Authorization: [METHOD]
- Logging: [SCOPE]
D.3 AI-Specific Security
- Model protection: [MEASURES]
- Input/output security: [MEASURES]
- Adversarial defense: [MEASURES]
This AI Data Processing Agreement template is provided for informational purposes. Legal counsel review is strongly recommended.
Do more with Ezel
This free template is just the beginning. See how Ezel helps legal teams draft, research, and collaborate faster.
AI that drafts while you watch
Tell the AI what you need and watch your document transform in real-time. No more copy-pasting between tools or manually formatting changes.
- Natural language commands: "Add a force majeure clause"
- Context-aware suggestions based on document type
- Real-time streaming shows edits as they happen
- Milestone tracking and version comparison
Research and draft in one conversation
Ask questions, attach documents, and get answers grounded in case law. Link chats to matters so the AI remembers your context.
- Pull statutes, case law, and secondary sources
- Attach and analyze contracts mid-conversation
- Link chats to matters for automatic context
- Your data never trains AI models
Search like you think
Describe your legal question in plain English. Filter by jurisdiction, date, and court level. Read full opinions without leaving Ezel.
- All 50 states plus federal courts
- Natural language queries - no boolean syntax
- Citation analysis and network exploration
- Copy quotes with automatic citation generation
Ready to transform your legal workflow?
Join legal teams using Ezel to draft documents, research case law, and organize matters — all in one workspace.