Skip to main content

Hello. It looks like you’re using an ad blocker that may prevent our website from working properly. To receive the best experience possible, please make sure any ad blockers are switched off, or add https://experience.tinypass.com to your trusted sites, and refresh the page.

If you have any questions or need help you can email us.

Oh look – Mark Zuckerberg’s trying to be cuddly

After the “perv glasses” disaster, the Meta boss is back with a new, cuter idea. And judging by the US response, it seems to be working. For now

Meta is betting that making AI cute and unthreatening will make us forget what it’s doing with all our information. Image: TNW/Getty

Everyone hates Meta, right? That awful company that’s given your kid ADHD (probably not) and via Facebook has turned your uncle into a spittle-flecked, bally-clad racist, that caused Brexit (maybe) and got Trump elected (hmmm) and played a significant part in fomenting Actual Genocide in Myanmar (this one happened). Except, well, look how adorable its new AI toy is!

Meta recently revealed its “Charm” AI product, a little plastic puck with a screen which, when released in time for Christmas, will act as a voice-activated controller for its AI assistant Muse. This has been greeted with the sort of press it seemed impossible for the widely-despised business to hope for in 2026. 

“Cute!” is the term that appears over and over again in coverage. Charm is palm-sized with rounded corners, a pleasing form-function, and comes complete with a cute cartoon mascot the company is choosing to call “Jolly”. In fairness, Jolly does look kind-of cute, a dumpy little creature with a vague “Inuit avocado” vibe to him and a bland, blank gaze onto which anyone can project anything they like.

If Jolly’s not your vibe, there are other options including an anthropomorphic mango, a labradoodle and a piece of toast wearing sunglasses. Tell Meta the sort of avatar that will make you think it’s an unthreatening friend and Meta will affect the demeanour.

Jolly will act as an anthropomorphised portal into the world of agentic AI; the idea is that users will unlock the device with a thumbprint and then speak directly to Jolly to issue instructions to their personal Muse agent which, linked to a user’s Meta accounts and email and other online identities, will be able to go and do stuff out there in the real digital world.

Objectively it’s a smart design decision from a company whose products have traditionally tended towards the charmless (see their “perv glasses”, for example) and whose reputational stock might best be described as “floating somewhere near the bottom of the Mariana trench”. 

As multiple commenters have pointed out, this latest offering hits very much at the heart of a contemporary trend for younger generations in particular to collect charms with which to adorn themselves and their belongings. Jolly is very much downstream of the Labubu / Jellycat obsession of recent years.

We’re also seeing an AI-driven growth in interest and appetite for what cultural theorist Jay Springett has come to refer to as “little guys”, digital creatures which live on our desktops or in our devices, a strange hybrid of pet and toy, companion and servant, experiment and game. 

Little Guys have been around for decades – the canonical early example is a game called Little Computer People from 1987, that let you create a Little Guy inside your ZX Spectrum to observe and interact with, but you can trace this through the Tamagotchi craze of the 1990s and to later iterations of the concept such as The Sims. 

The difference, of course, is that while those were toys that couldn’t escape the device on which they were housed, the Charm, and Jolly, is very much a tool. While not available in the UK (yet), the Muse AI agent, for which the Charm is effectively a controller and Jolly a relatable “face”, is the first real non-geek expression of the coming world in which AI takes real-world action.

What that means in practice is that you will be able to ask Jolly to do stuff for you. Use-cases are varied, but Meta is championing its agentic systems as a labour-saving technology which will free users up to Live More Life (streamed on pervert glasses, natch). 

Give it enough access and it will book tickets, cancel subscriptions, answer emails, make phone calls, plan and schedule nights out and even do your shopping, except on Amazon, which recently blocked the Meta AI from shopping on its site because, well, AI agents don’t window-shop or look at ads, and Amazon might just have some ambitions in the AI space itself. 

Early feedback from users of Muse in the US has been remarkably positive – millions of app downloads, and countless articles in which journalists seem to concur that, yes, this does actually work properly and is actually useful. Meta’s so confident about its utility that it’s pitching the AI explicitly as a money-saving device, claiming it can reduce monthly expenses by $1k by haggling over the phone with insurance companies for better deals, unsubscribing from that streaming service you never use, and sifting through the web for discounts.

The one caveat to all that is that you have to trust Meta with an agent that can act across your email, calendar, shopping, payments and assorted other corners of your online life – and why would anyone do that, given the company’s questionable track record when it comes to user data and ethics in general?

Which brings us back to the Charm device, and to Jolly. It’s a lot easier to believe that the tech’s benign when it comes packaged in an unthreatening little box that feels good in your hand, and when you’re issuing instructions to a lovably-proportioned, smiley little dude who’s always happy to help. Meta might ride roughshod over data-protection protocols and sell your information to the highest bidder – but Jolly? Jolly’s your friend!

It’s also a neat way to promote the always-on, always-here, all-knowing Meta AI to people who might be a little creeped-out, say, by the prospect of wearing the nonce goggles (sorry, Meta Glasses). Don’t like the face-mounted surveillance stuff? Have the MeTamagotchi instead (it also has cameras, but, well, it’s cute). It’s the benign face of the AI panopticon!

There’s an odd, ironic contrast between the current crop of apocalyptic headlines born of “rogue agents” going around hacking thousands of websites and the same websites running headlines talking about how useful Muse is and the weird, protective feelings elicited by Jolly (look at his little face!) People do realise this is the same tech, right? 

Meanwhile, early reports show that Muse has already been accused of reading users’ private messages, attempting to hack their Gmail and nudging them to share passport information and bank account details. This doesn’t seem to be putting US users off just yet, as they continue to download the app in their hundreds of thousands. 

It suggests that Meta’s PR strategy of “just make it cute and unthreatening” is working rather well. It also shows that, whatever our misgivings about AI, its safety and its potential impact on the world, we’re willing to set them all aside for a few quid saved on subscriptions, a digital slave to organise the lads’ night out, and a cute little guy smiling back at us from a tiny screen, always ready to do our bidding. Just don’t think about what he’s doing with all that information he has about you, or who, fundamentally, he answers to. 

Hello. It looks like you’re using an ad blocker that may prevent our website from working properly. To receive the best experience possible, please make sure any ad blockers are switched off, or add https://experience.tinypass.com to your trusted sites, and refresh the page.

If you have any questions or need help you can email us.