Advertising
The "AI Info" Label on Facebook and Instagram: What Businesses Need to Know
Why Facebook and Instagram put "AI info" on posts and ads, when you must disclose AI yourself, why stripping the tags is a bad idea, and how to post without AI.
- Meta puts "AI info" on posts when it detects industry signals that an image was made or edited with AI, or when the poster discloses it.
- Fully AI-generated content keeps a visible label. Content that was only edited with AI shows the label in the post's menu.
- You must disclose photorealistic AI video and realistic AI audio yourself. Ads made or edited with AI are labelled too.
- Stripping the hidden tags from files is not a fix. Posting real photos and branded designs made without AI is.
A furniture shop posts a picture of a new sofa. Within minutes a customer comments: "Is this even real? It says AI info." The sofa is real, and so is the shop. But the picture was made with an AI tool, and now the first thing customers notice is a label that makes them doubt it.
Many businesses run into this, and some tools now promise to "remove" the label. This guide takes a different approach. It explains what the label means, what Meta and TikTok actually require (checked on their official pages in November 2026), why removing the tags is the wrong fix, and how to make posts that look professional without needing AI at all.
What the "AI info" label is
Meta announced in February 2024 that it would label AI images on Facebook, Instagram and Threads, and in April 2024 it introduced a general "Made with AI" label. On 1 July 2024 Meta said it was "updating the 'Made with AI' label to 'AI info' across our apps, which people can click for more information." Meta admitted its labels "weren't always aligned with people's expectations", pointing to "some content that included minor modifications using AI, such as retouching tools."
Then on 12 September 2024 Meta added: "For content that we detect was only modified or edited by AI tools, we are moving the 'AI info' label to the post's menu." TechCrunch reported the same day that the label would still be "prominently displayed under content that was fully generated by an AI prompt."
So today there are, roughly, two situations:
- A picture fully generated by AI from a prompt: a visible "AI info" label under the post.
- A real photo edited with AI tools, such as a removed background or extended edges: the label sits in the post's three-dot menu, where fewer people see it.
How Meta knows a picture involved AI
Meta says its labels are based on its "detection of industry-shared signals of AI images": markers that AI tools write into the files they make. In its February 2024 announcement, Meta's Nick Clegg said the company was building tools to identify "AI generated" information in the C2PA and IPTC technical standards, "so we can label images from Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock." Meta's own AI images also carry "IPTC metadata and invisible watermarks," and at first, photorealistic images made with Meta AI were labelled "Imagined with AI."
C2PA is an open technical standard behind what are called Content Credentials. The C2PA describes them as "a cryptographically secure way to capture and express content provenance": a record, inside the file, of where it came from and how it was made. Many AI tools now add these records automatically. That's why a picture can be labelled even if you never told anyone it was AI.
What you have to disclose yourself
For video and audio, Meta puts the responsibility on the poster. The February 2024 announcement says Meta will "require people to use this disclosure and label tool when they post organic content with a photorealistic video or realistic-sounding audio that was digitally created or altered," and that "we may apply penalties if they fail to do so."
For a business, this means:
- An AI-generated video of a "customer" talking about your product, or a cloned voice in a radio-style advert, must be disclosed with Meta's AI label tool when you post it.
- A cartoon-style explainer that nobody would mistake for real footage is a different case from a photorealistic one. If in doubt, disclose.
- Political and social issue adverts have their own rules. Meta says advertisers "have to disclose when they digitally create or alter a political or social issue ad in certain cases." Check Meta's advertising policies before running one.
How to add the label yourself
When you share a post or Reel on Instagram, go to the last screen before publishing and open the advanced or more-options settings. Look for the control to disclose AI-generated or AI-edited content and turn it on. The wording changes between app versions, so look for "AI label" or "AI info". After publishing, check that the label shows.
If you added it by mistake, open the post's editing options and check whether you can change it. Only remove it if it is genuinely wrong.
Adverts get AI labels too
In February 2025 Meta said that ads created or significantly edited with its own generative AI tools would carry AI labels. Ads with photorealistic AI-generated people would show the label next to the "Sponsored" tag.
In an update dated 1 June 2026, Meta said it is beginning to roll out an "About this ad" section to the three-dot menu of every ad, will "begin automatically detecting ads created or edited using third-party AI tools," and will "apply an 'AI info' label included in About this ad." Meta notes the experience may vary by region because of legal requirements.
In short, once this detection rolls out, using an outside AI tool to make your advert will no longer keep it unlabelled on Meta.
TikTok works the same way
If you also post on TikTok, the rules point in the same direction. TikTok states: "We require creators to label AIGC that shows realistic-appearing scenes or people." It also says: "We also label AIGC from certain other platforms by using Content Credentials", the same C2PA records Meta reads.
Why stripping the tags is the wrong fix
Search for this topic and you'll find tools that promise to "remove AI info" by wiping a file's hidden data before you upload it. It's technically possible. The C2PA's own FAQ says its records "are typically embedded in the asset" but "can be separated." So why not do it?
- It doesn't change what you must disclose. Meta's rule on realistic AI video and audio depends on what the content is, not on what's in the file. Removing the data and staying silent is exactly the failure to disclose that Meta says it may penalise.
- It may not last. The C2PA notes that invisible watermarks and fingerprinting can help find a credential again after it's removed, and Meta says its own AI images carry invisible watermarks. Detection keeps expanding: in June 2026 Meta said it would begin detecting ads made with third-party AI tools.
- It misleads customers. A customer who orders the sofa in an AI picture and receives one that looks different has a real complaint, label or no label.
The label isn't really the problem. The problem is using an AI picture where a customer expects a real one.
Step-by-step: posts that look professional without AI
1. Photograph your real products, people and premises
Your phone is enough. Use daylight from a window, a plain background (a white wall, a sheet, a wooden table), and hold the phone steady at product height. Take several shots and pick the sharpest. For services, photograph the work: a finished hairstyle, a repaired engine, a cleaned office, your team on the job. If a customer appears, ask their permission first.
2. Keep a small library
Build a folder of good photos: each product from two or three angles, your shopfront, your team, and finished jobs. Twenty strong photos will cover weeks of posts. Add new ones whenever stock changes.
3. Use branded designs for announcements
Not every post needs a photo. Opening hours, a price list, a holiday notice or "new stock this Friday" work well as a clean design in your brand colours, with your logo and one clear headline. A design made from your own logo, colours and text, with no AI image inside, gives Meta no AI signal to label.
4. If you use AI, use it where it can't mislead
AI pictures can be fine for mood images, illustrations of an idea, or seasonal backgrounds, where nobody would mistake them for your actual product. Expect the label, and don't fight it. When you post realistic AI video or audio, add the disclosure yourself.
5. Never use AI in place of product photos
A customer buying a dress, a car or a meal needs to see the real thing. If you don't yet have a good photo, post a branded design that says "New arrivals. Message us for photos" and send real pictures in the chat.
6. Check before you post
Before publishing, ask three questions. Does this picture show something a customer could buy, as it really is? If AI was involved, would a customer feel misled to find out later? Is it a realistic video or voice that I need to disclose? If the answers are yes, no and handled, post it.
| What you post | What Meta does (as of November 2026) | What to do |
|---|---|---|
| Your own photo, not edited with AI | No AI signal in the file to label | Best for products, premises and team |
| Branded design made from your logo, colours and text | No AI signal in the file to label | Best for announcements and prices |
| Real photo edited with AI tools | "AI info" may appear in the post's menu | Fine for small fixes. Don't change what the product looks like |
| Image fully generated by AI | Visible "AI info" label if detected | Use for ideas and moods, never as a product photo |
| Photorealistic AI video or realistic AI voice | You must disclose with Meta's AI label tool | Disclose every time |
| Ad made or edited with AI tools | Labelled if made or significantly edited with AI: next to "Sponsored" for photorealistic AI people, otherwise "AI info" in "About this ad" (Meta's June 2026 update, rolling out) | Expect the label. Keep the offer honest |
How Rumidza handles pictures
Rumidza is set up for the honest, professional look by default:
- Your own photos and videos come first. In the post planner, every draft post without a picture offers "Add my photo or video", and you can upload a video file directly as well as photos. The planner also suggests a photo you could easily take with your phone, such as your product, your premises or your team at work.
- Branded designs with no AI. One tap turns a post's caption into a design using your logo and brand colours. It's drawn from your own material, with no AI image inside.
- AI pictures are optional. They're off by default. If you switch them on in Settings, the app tells you plainly that Facebook, Instagram and TikTok show an "AI info" tag on them. Rumidza leaves the provenance data in those files instead of hiding it.
- Your files are kept for you. Finished pictures and videos are kept for 90 days. If you want everything kept, the "Keep my files" add-on costs $5 a month.
You can start Rumidza free and try the branded designs on your next post. If you also use AI to write your posts or blog, our guide to using AI content without breaking Google's rules covers the writing side. For a wider view, see AI marketing for small businesses.
The "AI info" label isn't a punishment. It's a signal to your customers. Use real photos for real products, clean designs for announcements, and AI only where it can't mislead, and the label stops being something to worry about.
Questions people ask
Why does my Instagram post say "AI info"?
Meta adds the label when it detects industry-standard signals that an image was made or edited with AI tools, or when the person posting discloses it. Since September 2024, content Meta detects was only edited with AI shows the label in the post's menu, while content fully generated by AI keeps a visible label.
Can I remove the AI info label from my post?
If you added the label yourself, you can check whether the post's editing options let you change it. If Meta added it because it detected AI signals in the file, there may be no setting to turn it off; if you think it's wrong, use any feedback or review option the app offers. Removing a file's provenance data does not remove your duty to disclose realistic AI video or audio, and it can mislead customers. The reliable way to avoid the label is to post your own photos and designs made without AI.
Do I have to label AI images on Facebook and Instagram?
Meta requires people to use its disclosure tool when they post organic content with photorealistic video or realistic-sounding audio that was digitally created or altered, and says it may apply penalties if they do not. For still images, Meta labels AI content it detects.
Do Meta ads get AI labels too?
Yes. Meta labels ads created or significantly edited with its generative AI tools, and in a June 2026 update said it would begin automatically detecting ads made or edited with third-party AI tools and show an AI info label in "About this ad".
Does the AI info label reduce my reach?
Meta's announcements do not say that an "AI info" label on its own reduces a post's reach. The bigger cost for a business is trust: customers may doubt a product shown in a labelled picture. Separately, Instagram announced on 31 August 2026 that creators who feature an entirely AI-generated person without using its "AI-generated profile" label could see their recommendation reach limited.
Is it bad for a business to use AI pictures?
Not necessarily. AI images can be useful for illustrations, backgrounds and ideas. The risk is using AI images in place of real product photos, which can mislead customers about what they will receive. Be open about AI use and keep real photos for real products.
Sources
- Meta — Labeling AI-Generated Images on Facebook, Instagram and Threads (Nick Clegg, 6 February 2024)
- Meta — Our Approach to Labeling AI-Generated Content and Manipulated Media (April 2024, with July and September 2024 updates)
- TechCrunch — Meta is making its AI info label less visible on content edited or modified by AI tools (12 September 2024)
- Meta — Expanding GenAI Transparency for Meta's Ads Products (February 2025, updated 1 June 2026)
- TikTok — Supporting responsible, transparent AI-generated content
- Outlook Business — Instagram's new AI rule: undisclosed AI-generated profiles could lose reach (1 September 2026)
- gradually.ai — Instagram AI label guide (updated 31 October 2026)
- C2PA — Frequently asked questions about Content Credentials