Meta’s newly launched AI image generator, Muse Image, has already attracted some legal scrutiny, with an Australian disputes lawyer warning the technology is likely to expose businesses to defamation claims and breaches of Australian consumer law if it is misused.
Released earlier this week by Meta Superintelligence Labs, Muse Image is available through the Meta AI app, Instagram Stories and WhatsApp.
The tool allows users to generate AI images using public Instagram accounts as visual references by simply @-mentioning a profile, without notifying the account holder.
The tool even lets people change the clothing they, or others, are wearing.
While Meta has told B&T the feature includes extensive safeguards designed to prevent harmful or defamatory content, legal experts have voiced their concerns online about how easily public images can be repurposed without explicit consent.
Charlotte Hale, a practice leader in disputes and litigation at LegalVision, told B&T the technology highlights the growing gap between rapidly advancing AI capabilities and Australia’s existing legal framework.
She warned that businesses could fall foul of laws face significant legal risks if they use AI-generated images that falsely imply endorsements or commercial relationships
“If businesses use these images in a way that makes consumers believe public figures are associated with them or have sponsored their products or services, there’s potential for misleading and deceptive conduct,” she said.
“We’ve already seen examples of AI-generated celebrity endorsements circulating online, with Harry and Megan’s recent Australia visit for example,” she said. “This technology makes it much easier for businesses to create content that could mislead consumers into believing there’s a genuine affiliation.”
She added that Meta’s enormous commercial ecosystem could amplify those risks.
“With so many businesses operating on Meta’s platforms, there’s a real possibility of celebrity images being used to promote products or services, making people believe those individuals support or use those businesses.”
This is not an academic question, either. Small businesses began using AI to make it seem that Prince Harry and Meghan Markle were visiting their establishments in social media posts earlier this year.
The Beachside Bakehouse in Victoria, Australia, went viral after sharing a highly convincing AI-generated photo of the couple on their outdoor patio with the caption, “Next round’s on you Haz.”
Nina’s Cucina in Melbourne, also shared an AI image of the royals eating at the café as a joke after seeing other hospitality venues do it. The Instagram post got more than 800,000 views.
Hale also warned the technology could increase defamation risks if AI-generated images depict individuals in false or compromising situations.
“We’re seeing AI images placing people into situations that never occurred,” she said.
“If those images lower someone’s reputation in the eyes of the community, there is the potential for defamation claims against the businesses using them, or potentially against Meta itself for facilitating that manipulation.”
Meta maintains Muse Image has been built with extensive safeguards.
“We tested Muse Image extensively before deployment and built in protections from training through to the product experience,” a Meta spokesperson told B&T.
“The model is designed to avoid generating harmful content, and both your request and the final image generated go through safety checks to help ensure content doesn’t violate our policies.”
The company said every image generated by Muse Image is automatically watermarked using its open-source Content Seal technology, allowing anyone to verify whether an image was created using the tool.
Meta also said users who repeatedly attempt to generate policy-violating content can be temporarily blocked from accessing Meta AI, while serious cases of abuse may be investigated further.
For advertisers, Meta said responsibility ultimately rests with brands.
“Our safeguards are designed to help prevent policy-violating content, however advertisers are responsible for reviewing and approving all AI-generated content before using it in any campaign or creative.”
Meta said Muse Image is not designed to generate images of private individuals from text prompts alone. Instead, users can generate images of themselves by uploading their own photos or create images based on public Instagram accounts by @-mentioning users aged 18 or older.
Users can opt out by disabling the “Allow people to reuse your content on Instagram and with AI features at Meta” setting under Instagram’s Sharing and Reuse preferences. The feature is limited to public adult accounts, with private and teen accounts excluded.
Despite those safeguards, Hale believes lawmakers and regulators in Australia need to “move quickly” as generative AI becomes more sophisticated.
“The law is evolving and needs to evolve to combat these huge technological changes that we’re seeing,” she said.
“A lot of the laws we have today simply didn’t anticipate technology developing in this way. There certainly need to be updates to better protect both individuals and businesses going forward, whether through reforms or new legislation.”
