AI Accuracy and Source Policy
Last updated: May 2026
This is a draft policy for DECIFER's early access period. It is written in plain English and is intended to be transparent, not as a substitute for formal legal advice. Final versions will be reviewed by legal counsel before commercial launch. Questions? Contact [email protected].
How DECIFER uses AI
AI models are a core part of the DECIFER platform. We use them to interpret structured data, map relationships between signals, and produce plain-language outputs. AI is the interpretation layer, not the source of truth.
Sources first
DECIFER outputs are grounded in structured, verified sources. For Decifer Trading, this includes market data feeds, regulatory announcements, and structured financial information. For Decifer Learning, this includes curriculum-linked academic content. AI models interpret this material; they do not generate it from scratch.
Confidence and uncertainty
Where DECIFER provides an AI-generated interpretation, it aims to indicate the confidence level and the sources used. Not every output will be equally certain. We are transparent about the boundaries of what the model knows.
AI limitations
- AI models can produce errors, even when trained on high-quality data. We work to minimise this through domain-specific training and source grounding, but no AI output should be treated as infallible.
- Models may reflect patterns in training data that are incomplete or outdated. Real-time data helps, but there will be edge cases.
- Plain-language summaries may simplify nuance. Users making important decisions should read the underlying sources, not rely solely on AI-generated summaries.
No hallucination by design
DECIFER is specifically designed to reduce AI hallucination by grounding outputs in verified source material before applying language model interpretation. We do not allow the model to speculate without a source basis.
Feedback and corrections
If you find an error in a DECIFER output, please report it to [email protected]. We take accuracy seriously and will investigate every report.
Third-party AI providers
DECIFER may use third-party AI infrastructure providers. These providers are selected for their reliability, security practices, and data handling standards. We do not share identifiable user data with AI providers for training purposes.
Note: This is a draft policy for the early access period. Final legal review is required before commercial launch.