Skip to content

CryptoMetric AI AI Transparency Notice — counsel-review draft

Draft explanation of machine-learning assistance, human control and content disclosure; legal classification remains open.

Draft status and classification

This notice was drafted on 20 September 2026 and is not effective. [PENDING documented EU AI Act role, system classification, provider/deployer allocation and Article 50 applicability for each feature].

How machine learning is used

CryptoMetric AI is designed to combine explainable rule-based strategy evidence with a machine-learning second opinion. The output may support a user-configured trading decision, but it does not know the user's complete circumstances and is not presented as individualized advice or a reliable prediction.

When a user is interacting with AI

Interfaces must identify machine-generated analysis in context when a user could reasonably mistake it for a human judgment. [PENDING verified inventory of AI-assisted features, models, providers, versions and the precise notices displayed at each interaction].

Limitations and meaningful interpretation

Model output can be inaccurate, unstable, biased by data, stale or confident in error. A score or explanation does not prove causation, suitability or future performance. Users should compare the rule evidence, risk-gate result, source data and exchange state rather than rely on the model output alone.

Human control and contestability

The user selects configurations, can remain in paper mode, can stop automation and retains independent exchange access. Material service decisions such as suspension or support escalation should provide an understandable reason and a contact route where law or fairness requires review. [PENDING human-review workflow and response owner].

AI-assisted public content

Educational or public-interest content that is generated or materially manipulated with AI must receive accountable editorial review and any disclosure required by law before publication. Article metadata should identify the human author or reviewer and review date; review does not turn uncertain analysis into fact.

Data and monitoring

[PENDING model input inventory, training and evaluation provenance, third-party model processing, retention, quality metrics, drift monitoring, incident criteria and change controls]. Personal-data processing is governed separately by the Privacy Notice and applicable data-protection law.

Questions and changes

Questions may be sent to the legal contact in the publisher snapshot once confirmed. Material changes to AI purpose, user impact, provider or required transparency will be recorded in a new dated version.

Primary sources