OneClub OS Data & AI Notice
Version 2026-09-28 · Effective September 28, 2026
What changed
- Adds section 5: every player coach is labelled as AI and says so when asked; the coaches are not therapy; how crisis messages are handled; break reminders for players under 18; and the parent or guardian confirmation for under-13 accounts. Adds the safety moderation check to the provider register and publishes the [AI crisis protocol](/ai-crisis-protocol).
- Creates one provider register for every OneClub OS AI integration and explains the current player-assistant and staff-assistance data boundaries.
- Adds the three staff coach chats (Wellness Coach for staff, Match & Training Analysis, Leadership & Culture Coach), which clubs switch on individually and which see players only as private labels.
Effective September 28, 2026. This notice supplements the Privacy Notice and identifies the AI providers and bounded uses currently configured in OneClub OS.
This page is informational and is not a second contract. It does not block ordinary access. AI use of health or wellness data has its own affirmative choice; accepting Terms or acknowledging this notice does not make that choice.
The listed player and staff AI features operate only within their current account, club, team, role, purpose, guardian, safeguarding and provider controls.
Questions or requests: legal@oneclubos.com.
1. Current AI services and providers
| OneClub OS feature | Provider | Information sent | Purpose |
|---|---|---|---|
| Wellness Coach | OpenAI API | The player’s message, up to eight recent permitted messages in that assistant thread, first name, age band, positions, playing level, goals and training days; a limited 14-day wellness, injury-signal or availability summary only with current health and separate AI-health permissions | Generate an answer for that signed-in player about their own readiness and recovery |
| Drill Coach | OpenAI API | The player’s message, up to eight recent messages in that assistant thread, first name, age band, positions, playing level, goals, training days, and relevant ratings and drill context | Generate a drill-directory answer for that signed-in player |
| Goal / IDP Coach | OpenAI API | The player’s message, up to eight recent messages in that assistant thread, first name, age band, positions, playing level, goals, training days, relevant ratings, open action-plan items, and development context | Generate a goal and individual-development answer for that signed-in player |
| Player coach safety check | OpenAI API (Moderation) | Each message sent to a player coach, with names and email addresses removed, and each answer a player coach writes before it is shown | Detect self-harm, sexual content involving minors, violence and harassment, so that a concern receives a fixed safety reply and authorized review instead of an AI answer |
| Staff Wellness Coach | OpenAI API | An authenticated staff member’s message, permitted recent thread context, their role and teams, and server-selected wellness, injury-signal and GPS-load summaries only for players in their authorized scope with current health and separate AI-health permissions; players use private labels such as P07, which are pseudonymous, not anonymous | Help staff spot who needs attention and plan load; never medical advice, diagnosis or return-to-play clearance |
| Staff Match & Training Analysis | OpenAI API | An authenticated staff member’s message, permitted recent thread context, role and teams, match results and team statistics; server-selected player load only for authorized players with applicable health and AI-health permissions, using pseudonymous player labels | Analyse matches, runs of matches and training load and draft short reports for staff |
| Staff Leadership & Culture Coach | OpenAI API | An authenticated staff member’s message, permitted recent thread context, role and teams, recent results and wellness participation aggregated only from authorized players with current health and AI-health permissions | Support leadership and team-culture thinking; safeguarding concerns are routed to the club’s safeguarding lead instead of being analysed |
| Ask ClubOS | OpenAI API | An authenticated staff member’s bounded product-help question and the non-personal product context needed to answer it | Answer product-navigation and product-use questions; player, health, private-message, video, and GPS content is blocked before provider egress |
| OneClubOS newsletter drafting | OpenAI API, only when enabled by a super admin with a dedicated connection | Curated public product facts and the super admin’s drafting prompt; no subscriber list, club records, player data, health data or private messages | Draft role-specific product tips and how-to guides; published links remain controlled by OneClubOS |
| Coaches Corner drafting | OpenAI API or Anthropic API, according to the configured tenant route | Approved non-personal coaching content and drafting instructions | Draft or summarize coaching material; player data is not authorized in this workflow |
2. What player assistants and staff coaches exclude
Player-assistant payloads exclude the player’s full name and email, guardian information, teammate records, clinical notes, raw GPS coordinates, photographs, and video. Identifiers in free text are scrubbed before the provider request. The server selects the player, tenant, assistant, and permitted context; the browser cannot select another player’s record.
Staff coach payloads never contain a player’s name, email, date of birth, contact details, guardian information, clinical notes, photographs or video. Each player appears only as a label that is valid for one answer, and OneClub OS restores the name after the answer returns, inside the club. The server applies the staff member’s own team scope, page permissions and the club’s feature switches; the browser cannot widen them. A club turns the staff coaches on separately from the player assistants.
No OneClub OS AI integration is authorized for targeted advertising, cross-club profiling, model training or fine-tuning on customer personal information. Outputs can be incomplete or wrong. Player assistants provide sporting guidance only and never medical advice, diagnosis, treatment, or return-to-play clearance.
3. Storage, retention, and provider controls
OneClub OS keeps player-assistant and staff-coach conversation history for 30 days and then removes it from active storage. Ask ClubOS help history is also limited to 30 days. What a staff member tells a staff coach about themselves (for example their role, teams, goals or preferred answer style) is kept as a short private note for that staff member and that coach only, never for players. The approved staff-drafting workflows keep only the resulting product records and operational evidence described in the Privacy Notice; they do not create a general provider-chat archive inside OneClub OS.
Player AI goal memories and preferences are separate from 30-day chat history. Inferred memories have an expiry and the daily cleanup removes expired active values; other memories persist until corrected, cleared or lawfully deleted. Staff coach memory is scoped to that staff member and coach and has no automatic 30-day expiry. A safety event and its review record, a guardian confirmation, AI acknowledgement and break-reminder record are separate evidence records; a fixed automatic deletion period is not currently configured for each. Support mode expires after 24 hours. These evidence records remain subject to applicable deletion requests, narrow lawful holds and the account or player lifecycle. Provider logs and backups are separate.
OpenAI states that API inputs and outputs are not used to train its models by default unless an organization opts in. Standard abuse-monitoring logs may retain content for up to 30 days, with endpoint, feature, security, legal, and approved data-control exceptions. Disabling response storage is not by itself proof of zero data retention. See OpenAI API data controls.
Anthropic states that API inputs and outputs are ordinarily deleted from its backend within 30 days, with exceptions for customer-controlled longer-retention features, an agreed zero-data-retention arrangement, usage-policy enforcement, safety requirements, and law. Content flagged by its safety systems and feedback can have longer periods. See Anthropic commercial data retention.
4. Choice, permissions, and rights
A person can use ordinary OneClub OS features without starting an optional AI chat. By tapping Start chatting, you acknowledge this informational notice and choose to start an optional AI coach chat under your current Terms. Health or wellness context requires both the underlying health permission and a separate affirmative AI-health choice, along with applicable guardian and account controls. Declining that choice does not remove ordinary access. Age and account routing are established during onboarding. Parent-operated routes have a separate parent confirmation; that attestation is not itself health consent.
You can choose not to chat, clear an available assistant thread, exercise the access, correction, deletion, and withdrawal rights described in the Privacy Notice and Consumer Health Data Privacy Policy, or contact legal@oneclubos.com. Withdrawing health-purpose permission blocks new Wellness Coach processing that depends on health information without blocking ordinary account access.
5. AI disclosure, safety, and limits of the player coaches
The Wellness Coach, Drill Coach and Goal / IDP Coach are AI, not people. Each is labelled “AI coach” in the inbox list and at the top of every conversation. Every conversation opens with the line “Your coach is an AI, not a person.” and an “About AI coaches” link that shows: “You’re chatting with an AI coach — not a person, doctor or therapist. For emergencies call 911; for crisis call or text 988.” The coaches are instructed to say that they are AI whenever they are asked and never to claim to be human.
AI coaches may not be suitable for some minors. They are soccer-development tools, not therapy, counselling or medical care. When emotional or mental-health topics arise, a coach answers briefly and points to a trusted adult, the club’s safeguarding contact and 988; it does not counsel. The Wellness Coach may give general recovery guidance with a reminder to stop if it hurts and see a qualified professional for pain or injury. No coach gives a diagnosis, injury treatment, rehabilitation plan or return-to-play clearance.
Messages suggesting self-harm, abuse, an eating disorder, bullying, violence or an urgent medical problem receive a fixed safety reply instead of an AI answer. The reply names 911, 988 and relevant trusted-adult or specialist support. Fixed keyword rules and OpenAI moderation check messages; moderation also checks answers before display. If a safety check fails, the coach does not answer. The separate concern record has a fingerprint rather than duplicated text, although the ordinary chat may retain the original message. Authorized human reviewers may act under safeguarding procedures and law; monitoring is not continuous. The AI crisis protocol gives details.
Eligible players aged 13 to 17 see a break reminder after a period of active AI chat set by the club (60 minutes unless the club sets another value between 15 minutes and 3 hours). Active time counts across all three coaches and restarts after 30 minutes without activity. The reminder repeats that the coach is an AI, not a person, and can be dismissed. OneClub OS keeps the running time and a record of each reminder shown or dismissed. AI chat for players under 13 is paused.
Players under 13 do not have their own account. AI chat for their player profile is paused because a parent login or confirmation cannot establish who is typing. Earlier chat history remains subject to its ordinary retention and deletion rules. A future child-use pathway requires parent notice, verifiable parental consent, parent controls and separate approval. Ordinary club access remains available.
The break-reminder records and the guardian confirmations are removed when the account or player record they belong to is deleted.