What is the AI – and what is it not?
What kind of AI operates here, how it is labeled, and how hallucinations are mitigated.
Smart Legal Pro utilizes language AI. To ensure you know what you are dealing with, here are the most important points presented transparently.
What the AI is
- An AI system built on top of leading third-party language models – not a single, self-trained model.
- Designed for information and preparation: making risks, clauses, and deadlines understandable and generating templates.
What the AI is not
- No legal advice and not a human with legal responsibility (limits of the AI).
- It does not make binding decisions with legal effect.
Labeling
The fact that you are working with an AI system is consistently indicated throughout the app (notice and disclaimer on every output). Documents generated by us (PDF/DOCX) additionally carry a machine-readable AI label in the file metadata – you can consciously deactivate this in the settings (AI labeling metadata). How we implement transparency in accordance with the EU AI Act can be read on the AI Transparency Page.
How errors are mitigated
- Masking of personal data before the model call.
- Legally curated templates ("playbooks") with sources and review status for the analysis.
- You can report incorrect outputs via the 👍/👎 feedback; the quality team reviews reports.
- Optionally, you can have sources added to a highlighted section (primary source search, subject to a fee, price shown in advance).
How reliable the results are is covered in how reliable is the AI.
Related topics: how reliable is the AI · limits of the AI · AI vs chatbot · the AI works visibly · is it secure · AI labeling metadata
Was this article helpful?
