AI & Responsible AI Notice
Idelio uses artificial intelligence to generate or transform content. This Notice explains the role of AI, important limitations, how content may be reviewed and what users must do before relying on Output. It supplements the Terms, Privacy Policy and Acceptable Use Policy.
1. You are interacting with AI
When you use a generation, assistant, recommendation, transformation, ranking or similar feature identified in the interface, your request is processed in whole or in part by an AI system. Output is generated or materially modified by AI; it is not created or verified by a human professional unless we expressly say otherwise.
AI features are identified clearly before or at the time of the first interaction where the AI nature is not obvious. Where required by applicable law, synthetic Output is marked in a machine-readable, detectable format to the extent technically feasible, and deepfakes or AI-generated public-interest content carry an appropriate clear disclosure. Users must preserve disclosures and provenance required by law.
2. Providers and routing
We may route a request among the AI providers listed at https://idelio.pro/legal/technology-vendor-register based on feature, availability, quality, safety, language, geography and cost. The selected provider may receive the Input, generated Output, technical identifiers and safety metadata necessary to return the result and protect its service.
Provider practices differ by plan, endpoint and feature. Some providers retain API content temporarily, scan it for prohibited use, allow limited authorized human review or retain flagged content longer. Zero-data-retention is not a general feature promise. See the Privacy Policy and Technology & Vendor Register for current details.
3. Training and product improvement
We configure commercial AI services to avoid use of ordinary API customer content to train shared models where such controls are available. We do not use identifiable Input or Output to train or fine-tune our general/shared AI models unless:
- a) you separately and voluntarily opt in after receiving a specific notice; or
- b) a Business User enters into a written agreement that expressly defines the data, purpose, rights, retention and safeguards.
Declining or withdrawing an optional training consent does not prevent ordinary use. Operational processing, abuse monitoring, debugging and deidentified performance statistics are not model training, although they are still governed by the Privacy Policy.
Submitting feedback through a control marked for model or product improvement may include the associated conversation or asset. The notice at submission explains what is included and asks for the required choice.
4. Human review
Authorized personnel of SFER LABS LLC or a provider may review limited content when necessary to:
- a) investigate suspected abuse, fraud, security incidents or policy violations;
- b) resolve a support request or technical failure;
- c) comply with legal obligations;
- d) perform controlled QA or product research under the Privacy Policy; or
- e) process content covered by a separate voluntary improvement program.
Access is limited by role, logged where appropriate and subject to confidentiality. Do not use the Service as a repository for secrets or regulated data unless the specific feature is approved for it.
5. What AI Output can and cannot promise
AI can hallucinate, omit context and reproduce common patterns. Output can be inaccurate, biased, unsafe, offensive, technically defective or similar to material generated for others. It may contain or resemble names, logos, characters, styles, facts or other material associated with third parties.
No automated safety, quality, similarity or trademark signal is complete. Unless we provide a separate signed professional service, we do not perform a comprehensive clearance search and do not guarantee originality, uniqueness, registrability, non-infringement or legal protection.
An assignment of any rights we hold does not prove that copyright, trademark, design or other rights exist. Laws on authorship and AI-generated works differ by jurisdiction and continue to evolve.
6. Your responsibilities
Before using Output, especially commercially, you must:
- a) review it for accuracy, quality, safety and suitability;
- b) perform appropriate trademark, domain, copyright, design and publicity-right searches;
- c) obtain licenses, permissions and professional advice where needed;
- d) verify claims, citations and factual statements independently;
- e) avoid high-impact or regulated decisions unless we expressly approve the use case;
- f) disclose AI involvement and preserve provenance where law or context requires it; and
- g) avoid misleading people into believing synthetic content is authentic human, newsworthy or documentary material.
You remain responsible for how you select, edit, combine, publish and use Output.
7. Prohibited and high-risk use
The Acceptable Use Policy prohibits unlawful surveillance, impersonation, non-consensual sexual content, child exploitation, deceptive deepfakes, fraud, malware, targeted political manipulation, unlawful biometric uses, rights violations and unapproved use in employment, housing, credit, insurance, education, healthcare, legal services or other high-impact decisions.
We may block a request or Output even where content is lawful if reasonably necessary for provider rules, platform safety or risk management. A blocked result is not a legal determination.
8. Reporting and appeal
Report suspected infringement, unsafe content, bias, an incorrect AI disclosure or a moderation decision to [email protected]. Include the project/output identifier, issue and evidence you are authorized to share. We may preserve relevant records while reviewing the report.
Where required or appropriate, a person not solely relying on the original automated signal reviews an appeal. We may remove, disable or restore content, restrict accounts, notify affected parties or authorities, and improve safeguards.
9. Changes
A provider substitution within an already disclosed AI function may be recorded only in the Technology & Vendor Register. A materially new training use, content-review practice, data category, high-risk use or disclosure model requires this Notice and the relevant user controls to be reviewed before activation.
AI providers and capabilities change. We maintain provider and retention details in the Technology & Vendor Register so operational updates do not require rewriting this Notice. We update this Notice when our purposes, user responsibilities or material AI practices change.
Questions: [email protected] or [email protected].