Legal
AI Literacy
Last updated: July 2026
1. Why this page exists
The EU AI Act requires providers and users of AI systems to have a basic understanding of how those systems work - their capabilities, their limitations, and the risks of relying on their output. This page explains, in plain language, how RHM Digital uses AI so you can use the platform responsibly.
2. What RHM Digital's AI does
RHM Digital sends test prompts to third-party AI models (ChatGPT, Claude, Gemini, Perplexity) to observe how each one talks about your brand, and uses AI to turn the resulting metrics into readable summaries and suggested actions. In every case:
- The AI models are simulating a real user's conversation - their answers are a sample, not a guarantee of what every real user sees.
- AI answers can change between runs, between providers, and over time as the underlying models update.
- AI-generated summaries and action suggestions are advisory. You decide whether to use them.
- No AI output from RHM Digital makes an automated decision about you, your business, or any third party.
3. Known limitations
- Hallucination: AI models can produce plausible-sounding but incorrect statements.
- Sampling variance: a single test prompt is one data point, not a census of every possible AI conversation.
- Provider drift: model providers update their systems without notice, which can shift scores over time.
- Language and locale bias inherent to the underlying models is inherited by our measurements.
4. What RHM Digital is not for
RHM Digital is a marketing and AI-visibility analytics tool. It is not designed, marketed, or intended for high-risk or prohibited AI use cases under the EU AI Act - for example recruitment/HR screening, credit scoring, medical diagnosis, legal or government decisions, education grading, biometric identification, social scoring, or emotion recognition. Do not use RHM Digital for these purposes.
5. How outputs are labeled
Content generated or summarized by AI is marked with an "AI-generated" label inside the product so it's always clear which parts of the interface reflect an AI model's output versus a calculated metric.
6. More detail
For the full list of AI systems we use and how we classify their risk level, see our AI disclaimer. For how we handle your data, see our data processing page.