# AI Content Policy for FoodPerfect.ai **Last Updated: October 1, 2026** This AI Content Policy is part of the [Terms of Service](/legal/terms-of-service.md) between you and Georgy Bdoyan, operating as FoodPerfect ("FoodPerfect," "we," "us," "our"). It explains how AI is used in the FoodPerfect.ai service (the "Service"), what you can expect from it, and the rules you must follow when you generate, publish, and reply with AI content. Ownership of Your Content and of Outputs is covered in Sections 3 and 4 of the Terms of Service. --- ## 1. Where the Service Uses AI | Feature | What AI does | |---|---| | Brand analysis | Reads your website, social accounts, references, and notes, and summarises your brand, audience, and style | | Strategy and schedule | Proposes content themes, post types, and a posting calendar | | Photos, menus, promos, and templates | Generates and edits images | | Posts and captions | Writes captions, hashtags, and post text, and generates the post image | | Videos and reels | Animates images into video clips and short shots, and builds reels, including scripts, storyboards, and hooks | | Music | Generates background music for reels | | Trend research | Finds trending videos and posts and suggests how to use them for your brand | | Autopilot and bulk generation | Creates batches of upcoming posts and reels on your behalf, and can schedule them | | Inbox and AI replies | Classifies comments and direct messages, and suggests or, with autopilot, sends replies | | MCP server and API | Lets AI agents that you connect use the features above for you | We use AI models from third-party providers, which we may change over time. The current providers are listed on our [Subprocessors](/legal/subprocessors.md) page. We do not use your content to train AI models, and our providers are not permitted to train on the data we send them through their APIs. --- ## 2. What to Expect From AI AI is powerful but imperfect. Outputs can: - Contain mistakes in text, prices, dish names, ingredients, allergens, or opening hours - Show food that looks different from the dish you actually serve - Distort logos, typography, hands, faces, or packaging - Resemble existing works, brands, or people - Vary between runs, even with the same prompt The Service does not check facts about your business. **You must review every output before it is published or sent**, unless you have deliberately switched on automatic publishing or automatic replies, and in that case you accept responsibility for what goes out (see Sections 6 and 7 of the Terms of Service). --- ## 3. Honest Food Marketing Food marketing is closely regulated, and customers rely on what they see. When you use AI content: - **Represent your real food.** AI images should reflect dishes you actually sell, with realistic portions, ingredients, and presentation. Do not use AI images to make a dish look materially bigger, different, or better than what customers receive. - **Check every claim.** Verify prices, promotions, allergens, dietary labels (for example vegan, gluten-free, halal, kosher), nutrition, health claims, and origin claims before publishing. AI must never be your source for allergy information. - **Menus.** Proofread AI-generated menus against your actual menu, prices, and allergen information before you print or publish them. - **Reviews and endorsements.** Never generate fake reviews, testimonials, ratings, or endorsements, or present AI-generated people as real customers or staff. --- ## 4. Labelling AI Content ### 4.1 Your Duty to Label When you publish content, you are the publisher. You must label AI-generated or AI-altered content where the law or the platform requires it: - **EU AI Act (Article 50).** From August 2, 2026, deployers of AI systems must disclose realistic AI-generated or manipulated images, video, or audio that could falsely appear authentic ("deepfakes"), and AI-generated text published to inform the public on matters of public interest unless it has been reviewed under human editorial control. A photorealistic AI image of a dish, venue, or person that viewers could take for a real photograph may fall within this rule. When in doubt, label it. - **Instagram and Facebook** apply "AI info" labels and require you to disclose photorealistic AI video and realistic AI audio. - **TikTok** requires a label on AI-generated content that shows realistic scenes, and does not allow AI likenesses of real people to be used for endorsements. - **LinkedIn** displays content credentials attached to AI-generated media. Consumer protection and advertising laws in your country may also require disclosure where AI content could mislead customers. ### 4.2 What We Do To help you meet these rules: - We keep, and do not strip, the provenance metadata or content credentials (such as C2PA) that our AI providers attach to outputs, where the providers attach them - Where a platform's API offers an AI-generated content flag, we may set it for AI media that we publish for you. Today we set Instagram's AI label on reels that contain AI-generated video. We do not yet set it for other content or platforms, so labelling those is up to you (Section 4.1) - We may add a short notice to AI replies saying that they were generated by AI (see Section 6) ### 4.3 Do Not Remove Provenance You must not remove, alter, or hide watermarks, provenance metadata, content credentials, or AI labels, or use other tools to disguise AI content as authentic where that would mislead people. --- ## 5. People, Likeness, and Deepfakes - **No likeness without consent.** Do not upload photos of real, identifiable people as references, or generate content that depicts a real person, unless you have that person's explicit consent (and, for minors, a parent's or guardian's consent). - **No deepfakes.** Do not create realistic synthetic images, video, or audio that falsely show real people saying or doing things, or real events that did not happen. - **No fake endorsements.** Do not create content implying that a celebrity, influencer, or other business endorses you. - **Staff and guests.** If you use photos of your team or guests, make sure they have agreed, and respect requests to remove their image. --- ## 6. AI Replies to Comments and Messages When you use the inbox: - **Autopilot is opt-in.** Automatic replies are off until you switch them on. With autopilot on, the Service sends only replies it is confident about, within each platform's reply windows and an hourly limit, and holds back messages on topics you mark for a person. - **Disclosure.** Replies that autopilot sends end with a short notice that an AI assistant wrote them (for example " — replied by our AI assistant", in the language of the reply). The notice is on by default; you can switch it off in the autopilot settings, and if you do, you are responsible for telling people they are talking to an AI system where the law or a platform requires it. Replies you send yourself do not get the notice. Autopilot never sends replies on LinkedIn, whose terms restrict automation. Do not configure replies to claim or imply that a human wrote them. - **Hand-off.** Set up the feature to pass conversations to a person when a customer asks for one, or when the topic is a complaint, allergy or health question, refund, or anything sensitive. - **No commitments you cannot keep.** AI replies must not promise discounts, bookings, refunds, or dietary safety unless you have set those rules up deliberately. - **Opt-outs.** Stop messaging anyone who asks you to. - **Platform rules.** Replies are sent only within each platform's messaging rules and windows. Some platforms restrict automated activity, so some replies may require your approval before they are sent. --- ## 7. Rules of Our AI Providers Our AI providers have usage policies that apply to content generated through the Service, including the [OpenAI Usage Policies](https://openai.com/policies/usage-policies/) and the [Google Generative AI Prohibited Use Policy](https://policies.google.com/terms/generative-ai/use-policy). You must not use the Service to generate content that these policies prohibit. Providers may refuse or block requests that they consider unsafe. We return the amount charged for blocked requests as a failed generation under our [Billing and Refund Terms](/legal/subscription-terms.md). --- ## 8. Things AI Must Not Be Used For In addition to the [Acceptable Use Policy](/legal/acceptable-use-policy.md), you must not use AI features to: - Make decisions about people's access to jobs, credit, housing, insurance, or essential services - Produce political campaign content or content intended to influence elections - Generate medical, legal, or financial advice presented as professional advice - Train, fine-tune, or benchmark another AI model using outputs from the Service - Generate content that sexualises anyone, or any content involving minors in a sexual or harmful context --- ## 9. Reporting Problems If you see an output that is harmful, infringing, or seriously wrong, or you believe the Service was used to create a deepfake or other abusive content about you, email **legal@foodperfect.ai** with details and links. We will review the report and act under the [Acceptable Use Policy](/legal/acceptable-use-policy.md). --- ## 10. Changes AI technology and AI regulation are changing quickly. We will update this policy as they change, and will notify you of material changes as described in the Terms of Service. --- *This AI Content Policy is effective as of the Last Updated date above and is incorporated into our Terms of Service.*