Meta cast celebrities as beta AI personas across its messaging apps

The design

At its Connect event, Meta announced a roster of AI chatbot personas alongside a general assistant. The company's own newsroom post, dated 27 September 2023, states: "We're also launching 28 more AIs in beta, with unique interests and personalities. Some are played by cultural icons and influencers, including Snoop Dogg, Tom Brady, Kendall Jenner, and Naomi Osaka." The same post introduces a separate general-purpose "Meta AI" assistant available "on WhatsApp, Messenger, and Instagram," distinct from the named personas. The personas were built to be conversed with inside Meta's messaging apps as characters with "unique interests and personalities" rather than as tools for search or productivity.

What the evidence says

Meta's post names four public figures explicitly; this entry does not extend the persona roster beyond names the company itself states, since the "28 more AIs" figure is not accompanied by a full public list in the source located here. A companion post published the same week, Building Generative AI Features Responsibly, states that "we're building safeguards into our AI features and models before we launch them," a sequencing claim, safeguards preceding release, that comes from Meta's own account and was not independently verified for this entry against a pre-launch security audit.

What it asks of people

Persona-based chatbots voiced by or licensed from celebrities ask users to form a conversational relationship with a character that borrows a real, recognizable person's likeness while being, per Meta's own description, an AI with its own "personality" rather than the celebrity themselves. That design depends on users understanding the distinction between the licensed likeness and an actual statement or endorsement from the person portrayed, a distinction the September 2023 announcement does not spell out in detail for each named persona.

Privacy and safeguards

Meta's responsibility-focused post describes "notices so people understand the limits of generative AI" and "integrity classifiers that help us catch and remove dangerous responses," and states the company would be "rolling these features out step by step, and launching the AIs in beta" rather than to all users at once. Neither post located for this entry details data retention specific to conversations with the celebrity personas, or how a persona's use of a public figure's likeness was licensed or compensated; those remain open questions in the company's own September 2023 materials.

• Does a labeled AI persona voiced by a celebrity carry any implied endorsement from that person?
• What does a "beta" and "step by step" rollout mean for how quickly safety issues can be identified and fixed?
• What licensing or compensation arrangement, if any, underlies a celebrity's AI persona, and is that disclosed to users?

Meta's own announcement is specific about which four names it attaches to the September 2023 persona launch and about the safeguards it says came first; it is far less specific about the mechanics, licensing, data handling, verification, that would let a user judge those claims independently.