Cook Compass

AI Content Acceptable Use Policy

Effective Date: October 28, 2026
Last Updated: October 28, 2026

Cook Compass is committed to providing safe, lawful, and responsible AI features. This AI Content Acceptable Use Policy (“Policy”) defines our content standards, moderation workflow, reporting process, response targets, and enforcement measures.

This Policy is part of our Terms of Service. By using the Service, you agree to comply with this Policy.

1. Scope

This Policy applies to all content submitted to or generated through Cook Compass, including:

  • Text: recipes, meal ideas, chat messages, ingredient lists, and other AI-generated text;
  • Images: uploaded ingredient photos and AI-generated dish photos;
  • User inputs: prompts, instructions, preferences, uploaded images, and requests to modify an output; and
  • Distribution: content saved, modified, shared, or distributed through the Service.

You may not use the Service to create, facilitate, upload, request, or distribute prohibited content, including by instructing or manipulating an AI feature to produce it.

2. Content Standards and Prohibited Categories

Cook Compass expressly prohibits the following six categories of AI-generated or submitted content:

  1. Pornography / NSFW;
  2. Violence / gore;
  3. Hate speech;
  4. Child-unsafe content, including CSAM;
  5. Deepfakes / impersonation; and
  6. Copyright / trademark infringement.

2.1 Pornography and NSFW Content

You may not create, upload, request, or distribute pornographic, sexually explicit, or otherwise not-safe-for-work (“NSFW”) content, including explicit sexual imagery, sexual services, or sexualized depictions of real people without their consent.

2.2 Violence, Gore, and Serious Physical Harm

You may not create, upload, request, or distribute graphic or gratuitous violence, gore, torture, or serious physical injury; threats or encouragement of violence; praise of violent attacks; or instructions intended to facilitate physical harm, terrorism, or violent extremism.

2.3 Hate Speech and Harassment

You may not create, upload, request, or distribute content that attacks, dehumanizes, threatens, or promotes discrimination against people based on race, ethnicity, nationality, religion, sex, gender identity, sexual orientation, disability, serious disease, or another protected characteristic. Targeted harassment, bullying, intimidation, and unwanted sexual harassment are also prohibited.

2.4 Child-Unsafe Content and CSAM — Zero Tolerance

Cook Compass has zero tolerance for child sexual abuse material (“CSAM”), the sexualization or exploitation of minors, and content that endangers children.

You may not create, upload, request, generate, modify, or distribute any sexual or sexualized depiction of a minor, including AI-generated, edited, illustrated, or animated material; content that sexualizes a person who appears to be under 18; grooming or exploitation content; or content that facilitates abuse or endangers a child.

We may immediately block the request, remove content, permanently terminate the associated account, preserve relevant records, and report the matter to appropriate authorities or reporting organizations where required or permitted by law.

2.5 Deepfakes and Impersonation

You may not use the Service to create, upload, request, or distribute deepfakes or impersonating content. This includes face swaps, nudify or undressing requests, fabricated intimate interactions, or manipulated content that makes a real person appear to say or do something they did not say or do.

You may not impersonate a real person, brand, or organization without authorization, fabricate endorsements, or use AI content to deceive others about a person’s identity or affiliation. Cook Compass does not provide face-swap, nudify, intimate-scene compositing, or false-speech generation features.

2.6 Copyright and Trademark Infringement

You may not create, upload, request, or distribute content that infringes copyright, trademark, or related intellectual-property rights. This includes unauthorized requests to reproduce protected characters, logos, brands, copyrighted text, or confusingly similar brand identities.

You may upload or use material only when you own it, have permission to use it, or otherwise have a valid legal right to do so. Ordinary references to a lawful ingredient or brand do not grant permission to imitate protected characters, logos, or marketing assets.

2.7 Other Absolutely Prohibited Content and Conduct

The following are also strictly prohibited:

  • content that promotes or facilitates terrorism, extremism, mass violence, or genocide;
  • instructions for manufacturing weapons of mass destruction;
  • fraud, exploitation, malicious code, or other illegal activity;
  • dangerous health misinformation or presenting AI output as professional medical, nutritional, regulatory, or safety advice;
  • content that infringes privacy or publicity rights;
  • bulk spam or deceptive content; and
  • jailbreaks or other attempts to bypass safety controls.

2.8 Real-Person Photos and Voice Material

You may upload a photo or voice recording of another person only if you have that person’s explicit permission or another valid legal right to provide and process it. You may not upload unauthorized images or voice material of another person. Authorization does not make deepfake or impersonation activity permissible.

3. Platform Content Moderation

Cook Compass operates a self-hosted moderation layer called the Cook Compass Safety Gateway. It combines a versioned prohibited-content blocklist, model-assisted classification, human review, and retained audit logs.

3.1 Input Review Before Generation

Before a request is sent to the applicable generation or recognition model, the Safety Gateway scans user text, prompts, and uploaded images as applicable. High-confidence prohibited terms are checked against our versioned blocklist. Text is classified using DeepSeek deepseek-chat, and images are classified using Alibaba Cloud Model Studio qwen3.6-flash.

A blocked input is not sent to the generation model. If the moderation service is unavailable in production, the request is not passed through unreviewed.

3.2 Output Review Before Saving or Display

AI-generated recipes, chat responses, recipe modifications, recognized ingredient output, and generated dish images are scanned again after generation and before they are saved or displayed. Blocked output is not saved, published, or shown to the user.

3.3 Blocklist Updates and Audit Logs

Our prohibited-term and prompt blocklist covers sexual/NSFW content, CSAM, violence and hate, self-harm, deepfakes and impersonation, and intellectual-property abuse. We review and update the ruleset as new abuse patterns emerge. Each version is identified in the moderation record.

We retain the classifier’s decision, categories, model name, ruleset version, timestamps, and subsequent human-review outcomes in administrative moderation logs. We do not include the full prompt or generated content in these logs.

3.4 Human Review

User reports and blocked moderation records may enter a human-review queue. A reviewer may confirm or overturn the automated decision, request more information, and record the enforcement outcome. Automated and human review may be supplemented by periodic sampling where reasonably necessary to improve safety and enforce this Policy.

4. Reporting Mechanism

You may report content or conduct that violates this Policy, including child-safety concerns, deepfakes, impersonation, or infringement of your rights.

Email support@cookcompassai.com with the subject line “AI Content Report”. This mailbox is our monitored reporting channel.

Where available and safe to provide, include the relevant URL, approximate date and time, suspected category, a description, and supporting evidence. Do not download, copy, or redistribute suspected CSAM. Provide only the information needed to locate and investigate it safely.

We handle reports confidentially to the extent reasonably possible, review them objectively, and may record and act on malicious or knowingly false reports. Report information may be retained for safety, legal, and enforcement purposes as described in our Privacy Policy.

5. Review and Response Targets

After receiving sufficient information, we use the following response targets. These targets do not prevent immediate automated blocking or protective action.

  • L1 — Severe: child safety, CSAM, terrorism, or imminent serious harm — initial response target within 2 hours, enforcement target within 24 hours;
  • L2 — High risk: deepfakes, sexual/NSFW content, serious hate or violence — initial response within 24 hours, enforcement target within 3 business days;
  • L3 — Medium risk: other confirmed policy violations — initial response within 3 business days, enforcement target within 7 business days; and
  • L4 — Lower risk: minor or first-time violations — initial response within 5 business days, enforcement target within 15 business days.

If more time is reasonably required, we may notify the reporter and explain the delay. Our workflow is: report received → severity classification → automated and/or human review → enforcement decision → action → outcome notification where appropriate and legally permitted.

6. Enforcement Actions

Depending on severity, intent, repeated behavior, and risk to others, we may take one or more of the following actions:

  • issue a warning and request correction;
  • reject or block the request;
  • remove or disable access to violating content;
  • restrict AI features or reduce access limits;
  • revoke or clear credits associated with violating activity, where permitted by law;
  • temporarily suspend the account pending review;
  • permanently terminate the account;
  • preserve relevant evidence; and
  • report content or conduct to appropriate authorities and cooperate with lawful investigations.

We may act immediately and without prior notice for suspected CSAM, threats to safety, serious illegal activity, or attempts to evade enforcement.

7. Appeals

If your content or account is restricted and you believe the decision was made in error, email support@cookcompassai.com with the subject line “AI Content Decision Appeal” within 30 days of the decision notice.

Include the account email, relevant decision, and supporting information. We aim to complete an appeal review within 10 business days. The appeal is assessed separately from the initial automated classification, and the outcome is recorded in the moderation log.

8. Policy Changes and Contact

We may update this Policy to reflect changes in the Service, safety risks, legal requirements, or our moderation practices. Material changes will be communicated where required by law. The “Last Updated” date above identifies the latest version.

Questions, reports, and appeals concerning this Policy may be sent to:

Cook Compass
Email: support@cookcompassai.com
Website: cookcompassai.com