Cute Faces Like Muse Charm Are a Trojan Horse for Exploitative AI

A victims' rights attorney is comparing Meta's new AI assistant, Muse Charm, to a Trojan horse — a deliberately disarming gadget that invites users to hand over the most intimate records of their lives.

Carrie Goldberg, author of "Nobody's Victim," writes in a Los Angeles Times opinion column that Muse Charm is a keychain-size device resembling "a distant cousin of Furby or a Pixar character." Its soft, childlike appearance suggests nothing about the power of the technology behind it, she argues, or the consequences of giving that technology access to private lives.

"A monster is still a monster, even when it's been carefully designed to make us forget what it's capable of," Goldberg writes.

What does Muse Charm ask users to share?

The device encourages users to connect email, messages, photographs and financial accounts. AI assistants promise to do more as they learn more about users. Through that process, Goldberg argues, people create detailed records of their private lives with few guarantees about how the information will be protected.

The risk extends beyond the people who buy the product, she writes. Even people who never download or use the Muse software could have their private information exposed if someone else decides to connect their accounts. Privacy frameworks generally focus on the individual who agrees to share — but private communications involve multiple people, and one person's willingness to share should not override everyone else's reasonable expectations of privacy, the column argues.

Goldberg points to texts about an abusive relationship or an intimate photograph sent to a partner. "You may have consented to sharing that information with someone, but you didn't consent to sharing it with an AI company," she writes.

What precedents does the attorney cite?

Goldberg's firm represents survivors of sexual exploitation, stalking and technology-facilitated abuse. She says her firm has represented survivors whose intimate images were distributed online without consent, sometimes alongside identifying information that made it easier for strangers to find and harass them. Reclaiming control often takes years.

The pattern, she writes, extends to increasingly powerful technologies:

• Goldberg's firm is suing xAI, the company behind Grok, over allegations that its technology enabled the creation of sexualized images of a real person without consent.
• Flock's license-plate-reading camera network has been implicated in cases of police officers improperly tracking their former romantic partners.
• The firm has pursued a product-liability argument against Amazon over its sale of sodium nitrite, a chemical used in suicides. The lawsuits allege Amazon effectively sold suicide kits by recommending sodium nitrite alongside other products used to facilitate suicide, despite knowing the dangers.

How does the column frame the accountability question?

Companies developing AI know that predators, stalkers and abusers will look for ways to exploit their technology, Goldberg writes. Lawmakers and regulators have not kept up with emerging technological threats, but the legal principle that can hold AI companies accountable is well established: companies that design and distribute dangerous products can be held responsible when those products cause foreseeable harm.

She draws a historical parallel to tobacco companies using cartoon characters to appeal to younger consumers, vaping companies facing scrutiny over candy flavors and child-attracting designs, and alcohol companies relying on bright colors and playful packaging. With artificial intelligence, she writes, a similar dynamic is unfolding with remarkably little scrutiny.

Meta is not alone, the column notes. Goldberg points to Google's rounded letters and primary colors, which evoke children's alphabet blocks — a friendly identity for a company with enormous power over what people find online.

The stakes fall unevenly, she writes. Across decades of technology adoption, victims have too often been women, children and others whose privacy and safety were treated as secondary concerns.

Goldberg's closing argument is direct: it is easier to convince people to hand over their private information when the technology asking for it looks like a helpful friend — and behind the friendly branding sit extraordinarily powerful systems with access to increasingly intimate details of users' lives.