Electronic Frontiers Australia has issued an urgent warning to Australian social media users following the rollout of Meta’s latest AI tool, ‘Muse Image’. The feature routinely exploits public Instagram accounts by allowing anyone to co-opt and alter a person’s facial likeness using a simple text tag, without prior consent or subsequent notification.
EFA strongly condemns Meta’s choice by embracing “dark patterns” by deploying an aggressive “roach motel” UX model to manage user privacy instead of clear user notification and collection of express consent. By defaulting millions of public profiles into an AI-scraping ecosystem and hiding the opt-out mechanism behind layers of counter-intuitive account settings, Meta is explicitly capitalizing on user inertia rather than respecting individual personal data autonomy.
“Australian citizens should not wake up to find their faces, their children’s photos, and their personal creative outputs treated as free raw material for some random strangers’ generative AI experiments,” said John Pane, EFA Chair. “This rollout completely bypasses the core tenets of informed consent and Privacy by Design principles. It is simply another one of Meta’s ‘privacy breach mea culpas’ waiting to happen… in other words, wait for the well used Meta script sourced from their privacy infringing user experience playbook where overzealous engineers will be blamed. Make no mistake, this is a purposefully designed breach of privacy and not a product engineering error.”
EFA has in its initial examination identified several critical flaws inherent to the Muse design and architecture:
- Absolute Secrecy: Meta contractually abdicates its transparency obligations by refusing to notify users when their personal images are scraped and manipulated.
- Permanent Exposure: The opt-out mechanism is explicitly engineered to be non-retroactive, meaning any synthetic images generated during the default window remain active permanently.
Meta utilizes a textbook “Roach Motel” user experience (UX) dark pattern to manage user choices:
- Opt-In by Default: Instead of requesting affirmative consent, Meta automatically activates AI harvesting for every public account worldwide.
- Friction-Heavy Opt-Out: To revoke permission, a user must actively hunt through a hidden, multi-tiered hierarchy to find the on/off switch for this privacy invasive “feature”:
- Deceptive Interface Design: Users must manually toggle off separate, unlinked switches for Posts, Reels, and Audio.
- The Visual Interface: Uses nearly identical color states for “on” and “off,” exploiting user inertia and making accidental exposure highly probable.
EFA is calling on the Office of the Australian Information Commissioner (OAIC) to investigate whether Meta’s potential default processing of a person’s biometric facial data via Muse Image complies with Australian Privacy Principles. Until regulatory intervention happens, EFA urges all Australian Instagram users to immediately navigate to their Sharing and Reuse settings and manually disable all AI sharing toggles, or switch their accounts to private.
About Electronic Frontiers Australia:
Established in 1994, Electronic Frontiers Australia (EFA) is a national, membership-based non-profit organization promoting and protecting digital rights, online privacy, and civil liberties in the digital age.
Media Contact:
[email protected] or [email protected]
Image Credit: Unsplash/ Shutter Speed