The latest mechanism that Meta has introduced to Instagram requires no user consent button whatsoever — it activates automatically, and if you want to escape it, you yourself must dig into settings to disable it. This is the critical point driving Hollywood artists and millions of public Instagram account holders, including many Vietnamese Americans who use the platform to sell goods, advertise nail salons, restaurants, or promote entertainment careers, to suddenly become raw material for Meta's AI without ever being asked.
According to Variety, Meta has integrated a new image-generation model called Muse Image into the Instagram app, allowing anyone to tag a public account in a prompt to create AI images based on that account's content. Per The Hindu, this model was officially launched by Meta Superintelligence Labs on July 7, with the stated purpose of helping users design event invitations or develop creative concepts featuring other people's names. But the technical reality is far simpler than the marketing language suggests: a profile picture, a family photo, a product image that a pho restaurant owner posts publicly to advertise — all of it can be fed into an AI prompt by a stranger and transformed into a new image, while the account owner receives no notification whatsoever.
Default-allow is not a technical oversight — it is a calculated business choice that shifts the burden of action onto users rather than the company.
Meta's growth engine does not need anyone's permission
To understand why Meta chose a default-allow design instead of default-deny, one must look at the scale of the bet the company is placing on generative AI. According to TechCrunch, Meta's standalone Meta AI app launched in April last year reached 6.5 million downloads within just six weeks, and after Meta rolled out a new chatbot version, the app jumped from position 57 to rank 5 on the US App Store. Also per TechCrunch, Meta currently counts roughly 42% of the world's population as daily users of at least one of its platforms — an image, video, and interaction database of a scale that no competitor, not even OpenAI, can match.
At that scale, every percentage point of users who actively click to disable the feature is a percentage point of training content and interaction lost. Default-allow is not a technical oversight — it is a calculated business choice that shifts the burden of action onto users rather than the company. This is precisely what SAG-AFTRA, the union representing Hollywood actors and performers, pushed back on when it issued a statement that Meta is allowing anyone to use Instagram images to generate AI images without permission, while instructing members and all Instagram users how to disable the feature via the Sharing and reuse settings.
Who suffers most: artists, influencers, and Vietnamese-American family businesses
SAG-AFTRA is not the only union to speak out. According to Variety, CAA — one of Hollywood's largest artist representation firms — has directly opposed this policy to Meta and issued a public statement criticizing the way Meta forces creators to hunt down how to revoke their content rights, rather than asking first. This is a group with direct interests at stake and strong voice, so their pushback carries clear policy-advocacy weight, not a neutral technology assessment.
But the risk does not stop at the famous. Many Instagram accounts from the Vietnamese American community in the US — from nail salons, restaurants, real estate brokers, to singers and MCs working in the Vietnamese-language entertainment market in Little Saigon or San Jose — keep their accounts public because that is the only way to reach new customers. With the Muse Image mechanism, a shop owner's photo with a customer, a family photo posted on a holiday, or a food promotion image can all be tagged by anyone into an AI prompt to generate a different image — and according to The Hindu, Instagram has confirmed that users will not be notified when this happens, and the disable button has not even shown up for everyone yet.
The concern is not merely theoretical. Per The Hindu, many users are worried about the precedent set on the X platform, when that platform's AI tool Grok was once exploited to create fake sexual images from real photos of others. For a community like overseas Vietnamese, where family images, children, and personal reputation are tightly bound to small-business operations, the risk of images being misused for defamation or fraud is not some distant prospect — it is a hidden cost of having to use Instagram to exist in the digital economy.
The Sora lesson: is Meta repeating OpenAI's mistake?
Notably, Meta is not the first company to experiment with a default-allow model and pay the price. According to Variety, OpenAI's Sora video model previously operated under a similar default-allow mechanism and was shut down earlier this year following a wave of objections over image copyright and personal identity issues. This is a direct precedent, almost a forecast of what Meta might face.
The difference lies in scale and market position. OpenAI had to shut down Sora because the company lacks a native social network to keep users engaged another way — public pressure was enough to force a quick retreat. Meta is different: Instagram and Facebook are deeply embedded in the digital lives of nearly half the world's population, according to the 42% daily user figure TechCrunch cited. Pressure from actor unions or artist representation firms, though justified, is unlikely to force a company controlling dominant social communication infrastructure to shut down an entire feature the way OpenAI once did. Meta is more likely to choose a middle path — add notification layers, tighten a few terms of service — but maintain the default-allow mechanism, because that is the growth driver the company needs.
The Discover feed precedent: Meta has exposed sensitive information before, then quietly withdrawn it
Meta has not been without similar privacy incidents within its own AI ecosystem. According to TechCrunch, in the previous summer Meta had experimented with a feature called Discover feed on the Meta AI app, which publicly displayed AI conversation snippets that users had shared. The problem was that many users had inadvertently exposed sensitive personal information like home addresses, medical records, or private concerns in chat exchanges they thought only they could see. Meta subsequently had to remove the Discover feed entirely.
That retreat pattern — experiment widely, expose risks, then quietly remove when noticed — reveals Meta's product culture has a tendency to push privacy boundaries first, then adjust when there is enough pushback. Muse Image, with its similar default-allow mechanism, is following exactly that playbook. The question is not whether Meta will be forced to retreat, but how many images, how many identities will have been fed into the AI system before that happens.
What to do now, and its real limits
Both SAG-AFTRA and The Hindu outline how to disable the feature: go to Instagram Settings, select Sharing and reuse, then turn off permission for content to be used in AI features. This is a concrete step anyone can take in a few minutes. But let's be direct: this is individual self-defense, not a policy solution. It only works if users know this feature exists — and with tens of millions of Vietnamese Americans on Instagram, most of them small-business owners or older people who do not closely follow tech news, the likelihood they know to disable it themselves is very low.
This turns the Muse Image story into a test case for the entire industry: when generative AI platforms default to treating public content as free resources, the burden of self-protection falls entirely on individuals, while growth benefits flow to the company controlling the infrastructure. For the Vietnamese American community — which depends heavily on Instagram and Facebook to maintain small-customer networks, cross-continental family connection, and promote Vietnamese-language entertainment careers — the absence of strong policy advocacy voices like SAG-AFTRA or CAA makes this group among the most vulnerable to silent misuse of their images, with no one speaking on their behalf.
Conclusion: default is policy, not technique
Meta may be right that Muse Image is just a harmless creative tool. But the choice to design with default-allow, no notification, and shift the burden of disabling onto users, is an intentional policy decision — not a random byproduct of engineering. The Sora precedent shows the public can force an AI company to retreat, but Meta's own Discover feed precedent shows the company tends to choose partial-withdrawal solutions that keep the core mechanism intact. With nearly half of humanity as daily Meta users, the chance that the company will be forced to completely flip the default as OpenAI once did is low. Instagram users, especially small-business owners and Vietnamese American artists who depend on public accounts, should not wait for Meta to self-correct — the step to disable the feature in settings, however temporary a measure, remains the only concrete action currently in their hands.