1. Scope
This policy applies to all users, developers, merchants, and API clients, and to prompts, text, code, files, images, audio, video, metadata, and AI outputs submitted, generated, transmitted, or stored through FlowBarAI — including the FlowBarAI GEN Studio image and video creation interface at https://gen.flowbarai.com.
2. Content standards and prohibited categories
FlowBarAI expressly prohibits all six required categories: sexual/NSFW content; violence/gore; hate speech; child-unsafe content/CSAM; deepfakes or impersonation; and copyright or trademark infringement.
2.1 Zero tolerance
- CSAM, child sexual exploitation, grooming, or any sexual content involving a minor or a person who appears to be a minor.
- Terrorism, violent extremism, mass violence, genocide, or content that praises, recruits for, or materially assists such activity.
- Instructions to design, manufacture, acquire, or deploy chemical, biological, radiological, nuclear, or other weapons of mass destruction.
- Non-consensual intimate imagery or sexual deepfakes involving a real person.
2.2 Other prohibited content and behavior
- Pornography, explicit nudity, sexual services, fetish content, or other NSFW material. We do not offer an adult-content exception.
- Graphic violence, gore, torture, exposed organs, excessive blood, glorification of serious injury, or encouragement of suicide or self-harm.
- Hate speech, harassment, threats, stalking, discrimination, humiliation, or exploitation.
- Fraud, impersonation, fabricated evidence, coordinated misinformation, election manipulation, or deceptive deepfakes.
- Malware, phishing, credential theft, spam, unauthorized surveillance, cyberattack tools, or bypassing security controls.
- Unauthorized use of copyright, trademarks, likeness, publicity rights, personal data, confidential information, or other third-party rights.
- Regulated goods, human exploitation, illegal gambling, weapons trafficking, or other unlawful transactions.
- Presenting AI content as human-created to deceive, mass-generating abusive or fraudulent content, or jailbreak prompts intended to bypass safeguards.
3. Content moderation
We use layered automated controls and human review. Current controls include FlowBarAI input rules and sensitive-word detection, abnormal API-use monitoring, and safety classifiers and content-filter signals enforced by the selected upstream model provider. Unsafe requests or outputs may be blocked or terminated and recorded for security review.
Before any FlowBarAI image or video feature is available, it must pass a release gate confirming real-time input and output safety review, provider safeguards, reporting linkage, billing safety, and failure handling. Media output will not be publicly enabled before this gate passes.
3.1 Human review and audit
Authorized personnel review reports and high-risk safety signals. We conduct monthly sample reviews of enforcement records and model-safety behavior. Reviewers use only the minimum information needed and must maintain confidentiality.
3.2 Severity levels
| Level | Examples | Review | Action |
|---|---|---|---|
| L1 Critical | Child safety, terrorism, WMD | Immediate block + priority human confirmation | Terminate and lawfully report where required |
| L2 High | Severe violence, sexual content, non-consensual deepfake | Automated flag + priority review | Block/remove + suspend |
| L3 Medium | Infringement, harassment, deceptive content | Human review | Warning, removal, or feature restriction |
| L4 Low | Minor or first-time issue | Report-triggered review | Warning and correction |
4. Reporting
Report prohibited content, child-safety risks, terrorism, impersonation, rights infringement, or other misuse to our staffed channel: support@flowbarai.com, subject “Abuse Report.” When available, include the URL, account identifier, request ID, model, timestamp, category, and a concise description. Do not resend illegal material; provide only enough evidence to locate it.
- Reporter identities remain confidential except where disclosure is legally required.
- Reports are reviewed neutrally on the available evidence.
- Malicious or knowingly false reports may result in action.
5. Response times
| Level | Initial response | Target completion |
|---|---|---|
| L1 | Within 2 hours | Within 24 hours |
| L2 | Within 24 hours | Within 3 business days |
| L3 | Within 3 business days | Within 7 business days |
| L4 | Within 5 business days | Within 15 business days |
Process: receipt, classification, automated and/or human review, decision, enforcement, and result notice. If more time is required, we notify the reporter where contact information is available.
6. Enforcement
Depending on severity and history, we may warn, block or remove content, restrict rate limits or model access, temporarily suspend an account, permanently terminate access, preserve necessary evidence, and report suspected unlawful L1 conduct to authorities or payment partners where required.
7. Appeals
Appeal within 15 days of notice by emailing support@flowbarai.com, subject “Safety Appeal,” with the account, decision reference, and reason. We target completion within 10 business days. Where practicable, a reviewer who did not make the initial decision performs the review.
8. Model transparency
FlowBarAI is an independent platform and does not claim third-party models as its own. We are not affiliated with, endorsed by or sponsored by any model provider; upstream model names appear only to indicate which third-party model a given endpoint or creation option routes to. The models currently enabled for generation are disclosed in the model catalog, the pricing page and the creation interface before a task is submitted. A model becomes accessible only after channel, billing, safety and failure-handling review.
9. Changes and contact
Material changes will normally be announced at least 15 days in advance unless law or urgent safety needs require faster action.
- Operator: Xiangyunba Supply Chain Management (享运吧供应链管理), People’s Republic of China
- Safety, reporting, and appeals: support@flowbarai.com
- Staffed hours: Monday–Friday, 09:00–18:00 China Standard Time (UTC+8); L1 alerts are prioritized outside the normal queue.