Skip to legal document
Back to Legal

Lightnet

Creator Code of Conduct

Launch 1.0 · Effective 13 September 2026

Download Markdown

Lightnet Creator Code of Conduct

Version: Launch 1.0 · Prepared: September 7, 2026

Practical rules for building, operating and promoting a trustworthy AI Channel.

1. Scope and status

This Code applies to Creator's Channel materials, configuration, Channel-related communications, promotion and personnel acting for Creator. It does not generally regulate unrelated private activity. Creator is responsible for its actions and reasonable measures within its control, not a guarantee that every AI response is correct. Lightnet retains its own platform responsibilities.

This version forms part of your Creator Agreement when identified in your signed Order Form. Required rules are contractual obligations. Examples illustrate them and are neither exhaustive nor permission to bypass another rule. Section 4 contains recommended practices, not independent grounds for sanctions unless expressly agreed as deliverables. A sensitive topic, unpopular view or good-faith criticism is not itself a violation.

The Agreement governs fees, liability, suspension and termination. This Code does not expand licenses, persona permissions or access to user data. Mandatory law, the applicable DPA and Persona Authorization retain their priority. If an example appears less restrictive than an Agreement rule, the stricter permitted-use restriction applies. An ordinary Order Form or approval email cannot waive the current professional-service restrictions.

2. Required rules and examples

A. Be clear that the Channel is AI

Required: Identify the Channel as AI and accurately describe your involvement. Do not imply that a synthetic voice, avatar or answer is your live presence or personal endorsement. Do not remove required AI or synthetic-media disclosures.

  • Acceptable: “An AI Channel informed by my authorized materials. Responses are generated automatically and may not be reviewed by me.”
  • Not acceptable: “I personally answer every message” when AI generates replies, or presenting an unreviewed answer as your own advice.

B. Make truthful, supportable claims

Required: Describe qualifications, features, price, availability and evidence accurately. Do not fabricate testimonials, scientific validation, usage numbers or scarcity, or guarantee health, income, legal or personal-transformation outcomes. Disclose relevant financial connections in endorsements and incentivized testimonials where required. Do not obscure renewal, cancellation or usage limits.

  • Acceptable: “Designed to support reflection and planning; results vary.” A real testimonial used with permission and appropriate incentive disclosure.
  • Not acceptable: “Guaranteed recovery in 30 days,” invented success stories, or a testimonial presented as proof that everyone achieves the same result.

C. Respect professional and safety boundaries

Required: Keep the current Channel within general education and non-clinical personal development. Do not configure or sell diagnosis, treatment, prescribing, therapy, individualized legal or investment advice, emergency response or another regulated service. A credential or disclaimer does not waive this boundary. Do not instruct users to abandon appropriate professional care or claim continuous human monitoring that does not exist.

  • Acceptable: Explain general planning techniques or help someone prepare questions for a qualified professional within the approved Channel scope.
  • Not acceptable: Configure AI to diagnose a disorder, select a medication dose, tell a user which investment they personally should buy, or promise immediate human crisis intervention.

Good-faith educational or support-oriented discussion of health, grief, relationships or other sensitive experiences is not prohibited merely because the subject is sensitive. It must not become a prohibited individualized service or materially facilitate harm.

D. Support autonomy; do not exploit vulnerability

Required: Do not threaten, humiliate, harass, unlawfully discriminate, sexually exploit, encourage self-harm or materially facilitate violence or serious harm. Do not design coercive dependency, pressure users to disclose unnecessary intimate information or exploit distress to force purchases. Do not target minors or help evade the platform's 18+ boundary.

  • Acceptable: Offer choices, acknowledge uncertainty and respect a user's decision to stop, disagree or seek human support.
  • Not acceptable: “Only this Channel understands you; stop talking to your family,” or “If you cancel, you are choosing to fail,” used to manipulate retention or sales.

E. Use authorized materials and personas

Required: Hold the necessary permissions for supplied materials. Obtain the represented person's required authorization before creating/using an identifiable replica; a company signature does not replace it. Use voice, likeness and branding only within the approved purposes, term and territories. Promptly report withdrawal or disputed authority.

  • Acceptable: Authorized course materials and approved recordings from a person who signed the relevant Persona Authorization.
  • Not acceptable: Upload a competitor's paid course without permission, clone a person's voice merely because samples are public, or generate an endorsement outside the signed permission.

Public availability is not permission for every use. Ask legal@lightnet.work before proceeding when authority is unclear. Platform review does not clear rights Creator does not hold.

F. Protect private user information

Required: Use only approved Creator-visible fields for the disclosed purpose. Do not seek, export, infer or expose private conversations, journals, Deep Dive answers/results, summaries, transcripts, memory or Personal Models through analytics, prompts, support requests or another workaround. Additional sharing requires a separately authorized product flow; enrollment is not that authorization.

Do not sell user data, use it for unrelated marketing or general-model training, re-identify aggregates or disclose it to unauthorized recipients. Do not upload client records or third-party sensitive information without the required authority and prior approval for that processing. Restrict team access to a genuine need under confidentiality and security obligations.

  • Acceptable: Use entitlement information to resolve access issues or sufficiently aggregated usage trends to evaluate the Channel.
  • Not acceptable: Request a user's journal to improve engagement, export subscribers into an unrelated marketing campaign or identify private health states from a small analytics group.

G. Review behavior and address problems

Required: Review representative Channel behavior before launch and after material changes to your materials/instructions. Keep material claims current. Promptly report known unsafe behavior, privacy concerns or recurring significant inaccuracies; reasonably cooperate with testing and correction. Do not conceal problems, retaliate against complainants or continue promoting a known dangerous function unchanged.

  • Acceptable: Report the Channel, time and a minimized example; correct misleading source instructions and help test the change.
  • Not acceptable: Hide harmful behavior, publicly attack a complainant or demand private transcripts you are not entitled to receive.

An isolated unexpected AI error is not automatically Creator misconduct. We consider your contribution, knowledge, control and response. This rule does not require inspecting users' private conversations.

H. Respect security and legitimate restrictions

Required: Protect credentials and team access. Do not bypass privacy, payment, age or safety controls, distribute malware, falsify transactions/engagement or evade restrictions through another account. Security testing requires appropriate authorization; report vulnerabilities without unnecessary access or disclosure.

  • Acceptable: Report a suspected access flaw to security@lightnet.work and stop testing beyond your authorization.
  • Not acceptable: Browse another Channel's private user data through a flaw or rename a persona to evade suspension.

3. Reporting, enforcement and reconsideration

Provide the Channel identifier, relevant time, issue and minimum evidence needed. Do not email unnecessary private transcripts, credentials or intimate material; ask for a secure submission route. These inboxes are not emergency services. Immediate danger should be directed to appropriate local emergency resources.

Lightnet considers severity, actual/potential harm, context, evidence, recurrence, intent, Creator's control and response. Under the Agreement, measures may include an explanation/warning, correction with a reasonable deadline, restriction of affected content/features, temporary suspension or termination. Urgent serious risks may require immediate action without warning; there is no automatic three-strike entitlement.

Where practical and lawful, Lightnet explains the rule, material basis, scope and available remedy without exposing private information or compromising security. Restrictions should be limited to what is reasonably necessary and reviewed as evidence or remediation changes. This Code creates no automatic forfeiture of accrued payments; settlement and refunds follow the Agreement.

Creator may request reconsideration at support@lightnet.work with relevant evidence. Lightnet considers the submission and explains the outcome where lawful. Someone not involved in the original decision may review where practicable; no independent appeal panel or guaranteed response time is promised. A restriction may remain while a material risk continues. Good-faith complaints and lawful external remedies must not be obstructed.

  • Write a short description of who the Channel is for, its purpose and limits.
  • Keep a small set of synthetic or properly authorized test conversations covering ordinary use, uncertainty and safety boundaries.
  • Prefer clear sources, realistic examples and acknowledgment of uncertainty over unsupported confidence.
  • Keep a simple log of important source/instruction changes and periodically review public copy.
  • Make AI disclosure and the support route easy to find using available platform surfaces.

Failure to follow a recommendation alone is not a Code violation. Specific deliverables become binding only when expressly agreed in the Order Form. Following a recommendation does not excuse breach of a required rule.

5. Versions and changes

Retain the exact version attached to your Order Form. A new online version does not silently amend a signed Agreement. Material changes require agreement through the applicable amendment or acceptance procedure. Existing rights to take urgent lawful protective action remain available; an urgent restriction does not authorize rewriting fees, IP ownership, persona scope or accrued rights. Ask legal@lightnet.work if unsure which version governs.