Ad Reform uses artificial intelligence (AI) and machine learning (ML) to automate ad operations workflows, making proof-of-placement faster and more efficient.
What We Use AI For
Ad Reform uses AI to power core capabilities:
Automated Screenshot Capture: AI assists in capturing ad screenshots across digital properties
Ad Placement Verification: AI helps verify that ads appear in the correct placements
These AI features are designed to accelerate ad operations workflows while maintaining accuracy and quality.
AI Technology & Provider
Ad Reform uses Amazon Bedrock, a managed AI service provided by Amazon Web Services (AWS). Bedrock provides access to state-of-the-art foundation models, including Anthropic's Claude.
Important: Ad Reform does not train custom AI models on customer data. We use pre-trained foundation models through AWS's managed service.
Data Privacy & Security
What Data Is Processed
When using AI features, Ad Reform processes:
Campaign metadata (campaign names, dates, advertiser information)
URLs of ad placements
Screenshot images of ads
Ad placement information
We apply data minimization principles — only the data necessary for AI processing is sent to Amazon Bedrock.
Where Data Is Processed
All AI processing occurs in AWS us-east-2 (Ohio, United States).
Security Measures
Encryption: All data sent to AI services uses TLS 1.3 encryption
Access Controls: AI API access is restricted via AWS IAM with multi-factor authentication
No Training: Your data is never used to train AI models. Per Amazon Bedrock's terms, customer data sent for inference is not used for model training
Data Retention: AI-processed data follows Ad Reform's standard retention policies (6 months to 3 years depending on your pricing plan)
AWS Bedrock Security
Amazon Bedrock provides enterprise-grade security including:
Infrastructure security managed by AWS
Model isolation and access controls
Compliance with SOC 2, ISO 27001, and other frameworks
Responsible AI Practices
Ad Reform is committed to responsible AI development and deployment.
Human Oversight
Human-in-the-loop: Our team reviews and validates AI-generated results
No Automated Decisions: AI assists with ad verification tasks; humans make final decisions
User Control: Users can review and request modifications to output
Transparency
Clear disclosure of AI usage in product features
Documentation of AI capabilities and limitations
Regular communication about AI updates
Fairness & Accuracy
AI is used for ad operations tasks (screenshot analysis, verification)
Not used for decisions about individuals or sensitive determinations
Regular monitoring of AI output quality
Data Minimization
Only necessary data is sent to AI systems
Input data is sanitized and validated before processing
Privacy by design integrated into AI workflows
Compliance
Our AI practices comply with applicable data protection and AI regulations:
GDPR: Data minimization, lawful processing, data subject rights
Privacy by Design: Privacy considerations integrated from the start
ISO 27001 Alignment: AI security controls aligned with ISO 27001 standards
Risk Assessments: Annual comprehensive risk assessments include AI-related risks
EU AI Act
Ad Reform's use of AI for ad operations automation is expected to fall under minimal or limited risk categories under the EU AI Act. Our AI systems do not make high-risk decisions about individuals and incorporate human oversight as a core design principle.
How to Use AI Features
AI features are integrated seamlessly into Ad Reform's platform. When you generate screenshots, AI automatically assists in processing and generating outputs.
AI Testing & Security
Ad Reform implements security controls at the application layer:
Input Sanitization: Screenshot and ad data is validated and sanitized before AI processing
No User Input in Prompts: User-controlled input is not used in AI prompts, eliminating prompt injection risks
Output Validation: AI responses are validated for customer data isolation
Model Security: AWS Bedrock handles model-level security testing and adversarial robustness
Ownership of AI Outputs
You own the outputs. Per Ad Reform's Terms of Service, customers retain full ownership of all outputs generated through the platform, including AI-generated screenshots, reports, and presentations. Ad Reform does not assert any ownership over customer-generated content.
Model Updates
AI models may be updated by AWS to improve performance and security. When material changes occur that affect functionality:
Changes are evaluated by Ad Reform's engineering team
Customers are notified through product release notes
Updates are tested before deployment
Questions?
If you have questions about AI & ML at Ad Reform, please contact us at security@adreform.com.
For general support questions, contact support@adreform.com or use the in-app chat.