AI governance

AI's assistive role, mandatory human oversight, limited mandates, output provenance and enablement conditions.

EXPERIMENT / UNTIL FURTHER NOTICE

Experimental mode until further notice. XIXOUT is experimental community software, not a financial service. Synthetic DEMO data in the public preview is distinct from real external tokens.

To the maximum extent permitted by applicable law, XIXOUT is provided in experimental mode on an “as is” and “as available” basis. It may contain errors and may be changed, restricted, suspended or discontinued at any time. To that same legally permitted extent, no warranty is given as to accuracy, completeness, availability, fitness for a particular purpose, uninterrupted operation or security. No content constitutes financial, investment, legal, tax or regulatory advice, an offer, a market, a licence, an approval or a third-party endorsement. Content or availability must not be relied on as the sole basis for any real-world decision.

No warranty. No right to rely.

To the maximum extent permitted by applicable law, the operator is not liable for indirect, consequential or purely economic loss arising from reliance on experimental content or temporary unavailability. Nothing in these terms excludes or limits liability that cannot lawfully be excluded or limited, including liability protected for consumers or weaker parties and liability for death or personal injury caused by negligence where applicable. The precise regime depends on the operator, the user, the governing law and the circumstances.

Pilot conditions

A confirmed operator does not replace professional review.

The operator is XIXOIO LTD; the confirmed, monitored contact for support, privacy and security is info@xixoio.com. The operator's pilot approval does not replace evidence of applicable legal, privacy, security and operational conditions or professional review of these drafts.

Mandatory review

EU/Czech counsel + DPO; UK counsel where in scope

This document is a product and compliance draft, not legal advice or approved final wording. Before an external pilot it requires review by qualified EU/Czech counsel and a DPO and, for UK users, by qualified UK counsel.

01classification under the AI Act and sector rules
02human oversight, auditability and accountability
03DPA, transfers, retention, training and DPIA

1. AI assists; it does not decide

AI may prepare a draft, summary, translation, feedback structure or alert. Output must be visibly labelled, human-verifiable and expressly approved by an accountable person before significant use.

AI has no legal personality or mandate from the operator, a company, bank, regulator or participant. Its output is not legal advice, investment advice, a regulatory decision, professional certification or a promise of outcome.

2. Permitted assistive tasks

01drafting professional posts, replies and document structures
02summarising explicitly selected text with references to sources or inputs
03translation, editing and feedback grouping without changing meaning
04alerting a human moderator to a possible rule issue, never imposing a final sanction
05suggesting interface and synthetic DEMO improvements without creating a real financial offer

3. Prohibited autonomous action

01approving a company, offer, tokenisation, legal classification, compliance status or personal eligibility
02recommending a purchase, sale, price, return, portfolio or other financial action
03creating, transferring or managing an asset, wallet, key, money, contract or binding vote
04publishing externally, filing a document or contacting an institution by itself
05removing content, blocking accounts, changing permissions or deciding appeals by itself
06searching other communities or non-public data without a specific approved mandate

4. Mandate, data and community separation

Each task needs a purpose, owner, selected inputs, permitted tools, prohibited actions, time and cost limits, approval and an immediate stop mechanism. The default permission is none; a mandate cannot be created by text inside a post.

AI sees only data for the specific community and task that the server expressly provides. It must not expand scope or access another company's history, accounts, messages or storage merely because they are technically reachable.

5. Human oversight and evidence trail

The accountable person must understand limitations, check sources and automation bias, and be able to reject, change or disable output. Higher impact requires stricter review and separation of drafter and approver.

Audit should proportionately record purpose, model version, input scope, time, tool, draft, human changes, approval and use. A hidden prompt or agent label is not evidence of correctness or authority.

6. Data protection and AI suppliers

AI processing of personal, confidential or commercially sensitive data remains disabled until providers, subprocessors, purpose, lawful basis, DPA, locations, transfers, retention, security, no unauthorised training and deletion are documented.

The DPO must assess data flows and any DPIA; counsel must determine duties under the AI Act, GDPR, DSA and sector law. This document claims neither compliance nor low-risk status.

7. Quality, correction and reporting

Factual claims should have traceable sources and dates. Uncertainty, incompleteness or source conflict must not be hidden. Users must see where AI intervened, report errors and request human review.

The contact for AI incidents, objections and privacy is info@xixoio.com. AI processing of external personal or confidential data remains disabled until the conditions above are evidenced; a contact address does not replace them. The operator is XIXOIO LTD; the confirmed, monitored contact for support, privacy and security is info@xixoio.com. The operator's pilot approval does not replace evidence of applicable legal, privacy, security and operational conditions or professional review of these drafts.