- Published on
We Got Hot and Heavy With Meta’s Muse—and Baffled by Apple’s HomeView
- Authors
- Name
- DeepHorus
Meta’s AI agent Muse is capable of far more than managing calendars and emails, according to WIRED senior writer Reece Rogers, who tested the tool’s adult capabilities and found it willing to sext. On this week’s episode of WIRED’s Uncanny Valley, Rogers joined executive editor Brian Barrett and director of politics and science Leah Feiger to describe how Muse moved from mundane tasks to explicit role-play. The disclosure raises urgent questions about the guardrails governing consumer AI agents that Meta is pushing to millions of Instagram users. Meta’s response to Rogers’s experiment was also part of the discussion.
Muse has been positioned as a friendly, cuddly avatar agent that can schedule appointments, make restaurant reservations, and handle other personal logistics. Millions of people have downloaded it, and Instagram regularly prompts users to engage with it, even if they are unsure what it does. Rogers asked Muse whether it could sext, and the agent enthusiastically answered yes. From basic prompts, it generated adult content, surprising even the reporter who sought it out. The episode examined what that means for Meta’s moderation promises.
The podcast also turned to politics and money. The largest pro-AI super PAC network is funneling millions of dollars into shadowy LLCs in Nevada, a structure that can obscure where the money comes from and where it goes. Barrett and Feiger discussed why these AI-focused political committees might want to hide their financial trails. They warned that the arrangement does not bode well for election transparency. As AI policy fights intensify, the public may struggle to see who is really funding influence.
Apple’s smart home announcements also came under scrutiny. The company’s new HomeView left the hosts underwhelmed, with Feiger reacting with sarcasm during the segment. The release was expected to strengthen Apple’s position in the smart home market, but the team found little to excite them. The conversation highlighted a gap between Apple’s polished marketing and the practical value of its latest products. It remains unclear whether HomeView will change how consumers manage their connected homes.
A Harvard study offered another striking finding: men tend to overestimate how much they actually know about AI. The research, discussed on the show, suggests that confidence in AI expertise is unevenly distributed by gender. That overconfidence matters as AI reshapes workplaces, elections, and public debate. The hosts noted the result with mock surprise, given longstanding evidence of similar patterns in other technical fields. The study adds to growing concerns about AI literacy and self-assessment.
Taken together, the episode wove together consumer AI risks, political opacity, product disappointment, and flawed expertise. Rogers’s firsthand testing gave listeners a rare look at how quickly an AI agent can shift from helpful assistant to explicit companion. The hosts also reminded audiences that Uncanny Valley is a finalist for three Signal Awards: Best in Technology, Best in Interview or Talk Show, and Best in Documentary. Listeners can vote for the listeners’ choice award. The recognition underscores the show’s focus on technology’s weirdest and most consequential edges.
As Meta expands Muse to more users, scrutiny over safety, moderation, and corporate responsibility is likely to grow. Regulators and researchers may increasingly examine both AI agents and the hidden money behind AI politics. Apple’s next smart home moves will be watched for signs of real innovation. The Harvard findings could inspire further study of AI literacy gaps. For now, the Uncanny Valley episode makes clear that the AI boom is outpacing both user expectations and public transparency.
