Trust and safety

Automation & AI transparency

What the BubsBookings Safety Bot does, what data it uses, and where automated judgment stops.

Effective October 14, 2026

1. What is automated

BubsBookings currently uses in-house, rules-based systems that the product may call the Safety Bot. Content screening checks newly submitted customer service requests, provider quotes, messages, listing text, booking details, cancellation and refund reasons, reviews, disputes, reports, support requests, provider locations and policies, and team time-off reasons for selected patterns associated with profanity, harassment, hate, explicit sexual or exploitative content, credible threats, serious illegal transactions, and attempts to move marketplace-originated bookings or payments off platform.

Message screening combines multiple signals, including outside-payment language, contact details, external links, redirect phrases, and reasonable obfuscation, with quote, booking, and payment context. Contact information alone is not proof of circumvention. A phone number shared for arrival logistics after a paid booking is treated differently from a request to cancel and pay directly.

Listing screening also checks for selected financial-crime risk signals, including language suggesting payment without a genuine service, fake invoices, overpayment-and-refund arrangements, pass-through payments, cash-equivalent transfers, or concealment of funds. It can block a new or edited listing and flag an existing listing for administrator review.

Provider verification separately checks verified-email and agreement status, matching phone-number format, required business details, location, pricing, every active listing, and selected placeholder or suspicious language. When payout onboarding is connected, Stripe, not the Safety Bot-supplies the identity and payout-readiness status.

2. Inputs, outputs, and consequences

The systems use only the submitted fields, listing price, and account statuses described above. Content screening returns allow or block, a category, and a severity. Financial-risk screening returns a risk score, level, and non-public review reasons. Provider screening returns pass or request-changes, a score, and a checklist summary. A blocked submission is not posted; a provider profile that needs changes is not published through that submission. Neither system sets prices, accepts or denies bookings, ranks search results, decides disputes, files regulatory reports, or determines that a crime occurred.

3. Data records and outside AI

For most blocked content, BubsBookings stores the account, feature, category, severity, action, time, and a SHA-256 one-way fingerprint of the rejected text. For possible message-based marketplace circumvention, a restricted moderation record also preserves the attempted message, participants, conversation and booking context, rules and signals, action, and final review outcome so administrators can investigate, correct false positives, enforce the Terms, and handle appeals or disputes. These records are not delivered in the chat or exposed to other users. Provider screening stores its status, score, summary, and check time. These systems run with BubsBookings code and do not send private messages, booking text, or listings to a third-party generative-AI model for moderation. No outside AI model is being represented as the author of these rules.

4. Important limits

Rules can miss harmful or suspicious activity, misunderstand legitimate wording, slang, quotations, spelling, pricing, or context, and can be evaded. Listing text screening is only one control and does not replace Stripe's identity, sanctions, payment-risk, or transaction-monitoring controls. A matching, validly formatted phone number does not prove possession of that phone. A business-profile pass does not verify registration, licenses, insurance, credentials, service quality, or criminal history. Stripe identity status applies only to Stripe's onboarding requirements. None of these checks is a BubsBookings endorsement or background check, and a flag or block does not prove misconduct or a crime.

5. Human review and appeals

Administrators can review reports, disputes, listings, booking and payment records, and limited moderation-event information. A user who believes content or a provider profile was incorrectly blocked can contact support or email [email protected] with the affected feature, account email, and approximate time. BubsBookings may correct an error, explain a relevant rule, request revised content, or uphold the action. Material account or payout enforcement should consider the available context rather than relying only on an automated match.

6. Fairness and changes

The same content rules apply to customers and providers and are not designed to use race, color, religion, national origin, sex, disability, age, or other protected traits to rank or approve people. The Safety Bot does not make employment, housing, credit, education, insurance, health-care, or other legally significant eligibility decisions. Because language rules can affect groups differently, reports of repeated false positives should be reviewed. Material changes to the purpose, data, or consequences of these systems will be reflected on this page and, where required, presented for new consent.

7. Safety and emergencies

The Safety Bot is not monitored emergency response. If someone may be in immediate danger, call 911 or the appropriate local emergency service. Stop communicating with anyone who pressures you to leave the platform, send gift cards or cryptocurrency, reveal authentication codes, or provide unnecessary financial, identity, or health information. Use BubsBookings reporting and account-security tools for non-emergency concerns.

These pages describe the current BubsBookings service and policies. If a product screen appears inconsistent with them, contact support so BubsBookings can investigate and correct the issue.