AI Transparency

When you submit creative, a website, a brief, a competitor reference, connected performance context, or another supported marketing input, Critiqa can combine deterministic processing, stored evidence and third-party AI systems to produce structured analysis. Creative analysis currently includes 16 performance dimensions, while other product areas use task-specific evidence, scoring and diagnostic frameworks.

Scores are expressed as integers from 0–100 per dimension, with an overall weighted score and letter grade. These figures are estimates generated by an AI system, not assessments made by a human expert. They should be treated as directional guidance, not definitive verdicts.

Critiqa currently uses the following third-party AI services to process content:

Critiqa currently relies on third-party foundation and generation models alongside Critiqa's own orchestration, deterministic processing, evidence systems, evaluation logic and product controls. Legal roles and obligations can differ between upstream model providers and Critiqa as a provider or deployer of product features.

Under the EU AI Act (Regulation 2024/1689), AI systems are classified as prohibited, high-risk, limited risk, or minimal risk.

Critiqa evaluates advertising creative content. It does not make decisions about individuals' employment, creditworthiness, health, education, or legal status. It does not perform biometric identification. On this basis, Critiqa does not fall into a high-risk category under Annex III of the EU AI Act, and conformity assessments, CE marking, and technical documentation registers do not apply. This is Critiqa's own good-faith assessment, not a formal regulatory determination.

Article 50 transparency obligations have applied since 2 August 2026. They include disclosure duties for certain direct AI interactions and, in specified circumstances, marking or disclosure duties relating to artificially generated or manipulated content. Critiqa distinguishes obligations that fall on upstream model providers from obligations that apply to Critiqa's own product features and deployment role. Product labelling, provenance handling and this notice will continue to be updated as applicable guidance and implementation details evolve.

The Critiqa AI system:

• Analyses the visual and textual elements of advertising creative
• Generates numerical scores and written commentary based on learned patterns
• Compares creative elements against industry research benchmarks
• Suggests improvements to individual dimensions

The Critiqa AI system does not:

• Guarantee that following its recommendations will improve real-world ad performance
• Access connected advertising, analytics or publishing data unless you explicitly authorise the relevant integration and scope
• Make decisions about you as an individual
• Use your submitted content to train AI models
• Treat AI output as verified real-world performance unless supported by connected or otherwise grounded evidence

Right to explanation: You may contact us to request a general explanation of how a specific analysis score was derived. We can describe the scoring criteria and the weight applied to each dimension, but cannot provide a full trace of the underlying model's reasoning (which is opaque by nature of how LLMs work).

Right to contest: Critiqa's scores are advisory. You are never bound by them. Creative decisions remain entirely yours. If you believe a score is materially wrong, you can submit a support ticket and we will review it.

GDPR rights: Your personal data rights under GDPR (access, erasure, portability, rectification) apply separately and are described in our Privacy Policy.

Critiqa is designed to support human marketing and creative decision-making, not replace it. Scores and model-generated recommendations are estimates rather than final judgements. Human approval remains a deliberate boundary for consequential publishing, paid activation and intentional paid AI work where the product requires approval.

We review aggregate scoring patterns periodically to identify systematic errors or bias. If you encounter a score that seems clearly erroneous, report it via Support and it will be reviewed by a human.

AI creative scoring is an evolving field. Critiqa's benchmarks are informed by publicly available research (see methodology) but do not reflect every industry vertical, audience demographic, or cultural context equally.

Scores should be interpreted as one data point among many, alongside your own creative intuition, audience knowledge, and real-world test results.

For questions about our AI systems, this transparency notice, or to exercise your right to explanation: submit a support ticket or email [email protected].