This Schedule sets out the terms on which Provider uses artificial intelligence to validate purchases in Client's Programs. Capitalized terms not defined here have the meaning given in the GTC. In case of conflict with the GTC, this Schedule prevails for the AI Functions.
1. AI Functions
1.1 Scope. This Schedule covers the following functions of the Platform (the AI Functions):
- Receipt Scanning: AI-assisted reading of purchase receipts submitted by Campaign Participants, to extract the retailer, date, products and amounts needed to validate a purchase.
- Snap & Earn: AI-assisted recognition of a product or pack photographed by a Campaign Participant, to validate a purchase or consumption without a receipt.
- Image authenticity and fraud checks: the checks that support these two functions, such as live-capture controls and the detection of duplicate, manipulated or AI-generated images.
1.2 Client's Choice. Client selects the validation methods in the Order Form or, for Programs added under Section 2.3 of the GTC, in the change request. Where Client selects Receipt Scanning or Snap & Earn, it approves the use of the AI Functions for that Program on the terms of this Schedule. Where neither is selected, validation uses unique codes, QR codes or other methods, and no Client Materials are processed by the AI Functions.
2. How the AI Functions Work
2.1 Model Providers. The AI Functions combine Provider's own validation logic with AI models from the providers listed in Annex C of the DPA, used under their enterprise API terms, which exclude the use of inputs and outputs to train their models. Provider may move between the model providers listed there without further approval; new providers follow the sub-processor procedure of the DPA.
2.2 Data Sent. Only the image submitted by the Campaign Participant is sent to the model providers. No name, contact detail or other identifier of the Campaign Participant is transmitted with it.
2.3 Adaptive Validation. Receipt Scanning and Snap & Earn are adaptive: their accuracy on Client's products and retailers, and their resistance to fraud, improve as they process Client's Program data. Client agrees that Provider calibrates and optimises the AI Functions on that data during the Term, in accordance with Sections 7.2 and 7.3 of the GTC. Client Materials are not disclosed to other clients.
2.4 Human Oversight. Provider's operations team monitors validation outcomes, reviews flagged or contested submissions and can override an automated decision. Campaign Participants can contest a validation decision through the support channel of the Platform.
2.5 Transparency. The End-User Documentation informs Campaign Participants that AI assists the validation of their submissions.
3. Commitments
3.1 Regulatory Assessment. Provider has assessed that the AI Functions are not a prohibited practice, a high-risk AI system or a general-purpose AI model under Regulation (EU) 2024/1689 (the EU AI Act). Provider informs Client without undue delay if this assessment changes.
3.2 Security. The AI Functions are covered by the Security Measures (Schedule 2) and the DPA.
3.3 Accuracy. The AI Functions are designed and monitored to validate purchases reliably, but automated reading and recognition can contain errors. Validation thresholds and review rules are set with Client during Program design. Provider's liability for the AI Functions is governed by the GTC.
3.4 Information. Provider's Position Statement on the Use of Artificial Intelligence is available to Client on request, together with any further information Client reasonably needs to assess the AI Functions or to meet its own obligations under applicable AI laws.