WTFB uses artificial intelligence as part of its core functionality. This page explains where AI is used, what data is processed, and your rights — in line with the transparency obligations of the EU AI Act (Regulation (EU) 2024/1689), in particular Article 50.
1. Where AI is used
- Roast audits. When you submit a URL, an AI system fetches the page (where permitted by robots/policies), analyses the public HTML/screenshots, and produces findings, severity scores, and a verdict. Outputs are AI-generated content.
- Report scoring. Tester reports are evaluated by AI against quality rubrics (clarity, reproducibility, evidence). The score influences WTFx rewards.
- Translations. Some UI strings and reports are translated by AI between RU / EN / DE / UK / ES.
- Test-case generation and difficulty estimation. AI suggests test cases and estimates task complexity for project owners.
2. Disclosure of AI-generated content
Any roast audit, AI score, AI translation, or AI-generated test case shown to you is produced by an automated system, not by a human reviewer. We label such content where it appears (e.g. "AI audit", "AI score") so you can distinguish it from human-produced output.
3. Limitations and honest caveats
- AI outputs can be wrong, incomplete, or out of date. Treat them as a starting point, not as legal, medical, financial, accessibility, or security advice.
- The audit looks at what an AI can see from the public page. It does not replace a manual security review, a full accessibility audit, or human usability research.
- WTFx scoring is a heuristic. Disputes are reviewed by humans via the dispute flow.
4. Human oversight and dispute
Every AI scoring decision that affects rewards or visibility can be challenged. Use the dispute action on the affected report or contact [email protected]. A human reviews disputes and can override the AI result.
5. Data sent to AI providers
AI processing is performed via the Lovable AI Gateway and underlying model providers (e.g. OpenAI, Google Gemini). The data sent includes the public URL you submitted, fetched public page content, the text of your reports, and the prompts we construct. We do not knowingly send special-category personal data, payment data, or auth secrets to AI providers. See Subprocessors for the current list and Privacy for the legal basis.
6. No training on your data without consent
We do not authorise model providers to train their foundation models on your content via the gateway settings we use. If a provider changes that default, we will update this page and Subprocessors.
7. Prohibited uses
Do not use WTFB AI features to generate or distribute content that is illegal, defamatory, sexual content involving minors, real-person impersonation, targeted harassment, malware, or that violates third-party rights. See Acceptable Use.
8. Contact
Questions about AI use: [email protected].