Meta's Muse Image: a practical guide for brand marketing

Meta Superintelligence Labs released its first image model, Muse Image, on 7 July 2026, in the Meta AI app and on meta.ai. TechCrunch reports that it was developed under the codename Mango. It succeeds Emu, the model behind Meta's earlier Imagine feature, and it already powers new effects in Instagram Stories in the US. Meta says it is also coming to Advantage+ creative, the AI tools inside Ads Manager.

For any business producing a steady flow of social content, it is a tool to understand properly before the team starts leaning on it.

What the model is built to do

• Create images through conversation, starting from more than 30 suggested prompts.
• Plan before it draws. Alexandr Wang, who leads Meta's AI lab, called it agentic: it works with Muse Spark to reason, search the web and plan the layout before generating anything, and it can call tools to draw items such as charts and QR codes.
• Merge several reference photos into a single picture.
• Take edits drawn or annotated straight onto an image.
• Produce legible text, for example in infographics.
• Restore old photos and remove unwanted objects.
• Redesign rooms with real products from retailers or Facebook Marketplace.

Where you can use it

• In the Meta AI app and on meta.ai, since 7 July.
• In Instagram Stories, through more than 30 effects built on Muse Image, such as Night Flash and Disposable. These are US only.
• In WhatsApp chats with Meta AI, in a limited set of countries that Meta has not named.
• On Facebook and Messenger, which Meta says are coming.
• For developers, through Meta's model API since 25 August, at one US cent per image and up to 1,600 pixels, and through partners including fal, Runway and OpenRouter.

Meta has not confirmed Muse Image for the UAE or Saudi Arabia, nor that it handles Arabic. Try it in the Meta AI app on a local account before building plans around it, and test Arabic text in images before you depend on it.

How it compares

At launch, Meta pointed to LMArena, where people vote between images without knowing which model produced them. In early July, Muse Image came second in both text-to-image and image editing, behind OpenAI's GPT Image 2. The field has moved quickly since. By 21 September it had dropped to seventh in text-to-image and sixth in image editing on the same leaderboard, and Artificial Analysis also placed it seventh. Muse Video, previewed at the same time, ranked ninth in text-to-video. Anything still quoting the July positions is out of date.

Ranking matters less than fit. The strengths Meta highlights, laying out a picture and rendering text, suit social graphics and ad variations well. For the hero image of a campaign, a photographer or designer remains the safer choice.

The feature Meta pulled after three days

Muse Image launched with a feature that let anyone mention a public Instagram account and use that person's photos in a generated image. Public accounts were opted in by default, their owners were not notified, and the way to opt out was tucked inside a sharing menu. NBC News reported that it could be used to make deepfakes of both celebrities and ordinary people, and CAA and SAG-AFTRA objected. Meta removed the feature on 10 July, three days after launch, and admitted it had got it wrong. The model itself stayed.

Separately, a proposed class action filed in Chicago on 4 September 2026, as reported by a US law firm, alleges that Meta trained its image models, Muse Image among them, on Facebook and Instagram photos, including biometric data. Meta disputes the claim.

Watermarks, detection and labels

• Images from Muse Image carry an invisible watermark Meta calls Content Seal, which Meta says survives cropping, compression and screenshots. There is no visible mark.
• Engadget reported that Content Seal is proprietary and is not compatible with the C2PA industry standard or Google's SynthID.
• In a Reuters test of 40 images, Meta's own detection tool missed 55 percent once the images were heavily cropped. Meta said the tool is a preview.
• Meta labels AI content on Facebook and Instagram as AI info, based on industry signals or self-disclosure. Do not assume a Muse Image post will be labelled for you.

Ground rules before your brand uses it

• Disclose AI-generated images yourself by applying the AI label when you post, rather than trusting detection.
• Never generate real people, including your staff and customers, without their written consent.
• Keep logos and product photography real wherever accuracy matters. AI can alter a product's shape, colour or wording.
• Read the terms before publishing. Meta's terms for the Meta AI app do not clearly say whether its images may be used commercially, and images made with Meta's ad tools may only be used on Meta's platforms.
• Keep a record of prompts and outputs for anything you publish, in case a rights question arises later.
• If you run campaigns in Europe, transparency rules for AI content under the EU AI Act have applied since 2 August 2026.

Whatever made an image, check it in each Facebook and Instagram placement and crop it properly for every platform before it goes out.

ALXO works with AI-generated content and with real photography, so if you are deciding where Muse Image fits in your brand's content and where a proper shoot still pays off, we would be glad to talk.