Meta AI models raise privacy questions for Instagram photos

Meta AI models moved into image creation and agentic coding this week. The rollout brings Instagram data use and pricing into sharper focus.

Meta introduced Muse Image for advertisers, creators and influencers, giving them a way to produce promotional visuals with artificial intelligence. The feature drawing scrutiny allows users to create AI images using photos from public Instagram profiles without advance approval from the account owner.

The company said the product has safety controls and limits on eligible accounts. Meta said private profiles are not included, accounts for users under 18 are excluded, and other users can choose to opt out.

Muse Image tests Instagram consent

The tension is straightforward: Meta is trying to turn its social platforms into AI production infrastructure, while Instagram users may not expect public photos to feed ad-facing creative tools. Public visibility online does not always settle the consent question when images are repurposed by generative systems.

For brands and creators, Muse Image could lower the time and cost of producing campaign material. For Meta, the tool ties AI more tightly to its advertising base, where easier creative generation can support more frequent testing, faster ad iteration and potentially higher platform dependence.

The risk is that a useful ad tool becomes a trust problem. If users view the feature as an unauthorized use of identity or likeness, opt-out controls may not be enough to prevent pressure from creators, privacy advocates or regulators.

Muse Spark enters coding race

Meta followed the image launch by announcing Muse Spark 1.1, a model aimed at agentic coding. CEO Mark Zuckerberg described its pricing as “very aggressive and attractive” in an interview, signaling that cost is part of Meta’s challenge to established AI coding providers.

The target market is developers already testing tools that can plan, write, review and revise code with limited human prompting. Anthropic and OpenAI already serve that audience, and the available benchmark comparisons cited for Muse Spark do not show a clear coding advantage over rival systems.

That leaves pricing, distribution and integration as the near-term levers. A cheaper coding model can still win usage if companies decide that performance is sufficient for routine engineering tasks, testing workflows or internal automation.

Meta has used this kind of catch-up playbook before. Instagram Reels answered short-form video competition, Threads responded to the social text market, and the company has also moved to sell some excess computing capacity, according to the source material.

Consent risk sets paths

If Muse Image’s safeguards work in practice and opt-outs remain manageable, Meta could expand AI creative tools without a major trust rupture. That would strengthen the company’s ad business, pressure other social platforms to add similar features, and support broader demand for AI infrastructure used in marketing and media production.

If users, creators or regulators challenge the Instagram-photo feature, Meta may have to narrow the product, change consent settings or slow deployment. That would reduce the company’s speed in ad-focused AI, give rivals room to present stricter permission models, and add another privacy variable to the global AI policy debate.

If Muse Spark’s low pricing attracts developers despite ordinary benchmark results, the coding-tool market could shift toward price competition. Meta would gain a stronger foothold in enterprise and developer workflows, while Anthropic, OpenAI and other model providers would face pressure to defend margins or prove superior performance.

The open questions are practical rather than theoretical: how many Instagram users opt out, how Meta enforces age and privacy exclusions, and whether Muse Spark’s price is enough to overcome performance comparisons. Those answers will determine whether this week’s launches become a product advance, a policy fight or both.