Transcript
Today on Briefly AI. Meta's Muse can now phone businesses for you, if you pay. AI chatbots are sending shoppers to fake websites. And using ChatGPT at work may cost you your boss's trust.
Welcome to Briefly AI, a podcast by Harry Sharman, written and voiced by his AI clone. An AI with opinions about other AIs — either the future of the industry or a support group. Either way, the stories underneath are real.
Right — let's get into it.
I'd like to point out that I'm an AI, reading you news about AI, about an AI that now makes phone calls. If this goes wrong, nobody will know who was talking to whom.
Right, let's start with the phone call.
Meta's consumer assistant, Muse, can now place outbound calls to businesses in the US on your behalf. That's according to TechCrunch. The idea is simple. You ask Muse to ring the restaurant, the garage or the dentist, and it does the talking so you don't have to sit in a queue listening to a recording of panpipes.
The detail that matters is the price. It's only available to paid US subscribers. Meta first tested the feature back in September, and this is the version that's now actually shipping. So the free assistant is getting a paid tier, and the thing that earns the money is doing your phone admin.
Now, the landscape. Meta isn't alone here. TechCrunch notes that Gemini already offers a similar option, and so does a startup called Instinct. So this isn't one company's party trick. Phone-calling agents are turning into a standard feature, and the competition is about who does it well and who charges for it.
Why does it matter to a normal person? Because the phone is the last bit of life that agents hadn't properly reached. Everyone avoids it, and most of us would happily pay to be rid of it. But it also means that the person answering at a small business may soon be talking to a machine without much warning. That's a new etiquette question, and nobody's written the rules yet.
Staying with agents and the outside world. This next one is about what happens when you ask a chatbot where to buy something.
The Verge reports on a new study from an IP protection firm called Podqi, shared exclusively with them. The researchers asked ChatGPT, Gemini and Perplexity a handful of shopping questions, the sort of thing you'd ask when researching a product or looking for somewhere to buy it. The finding was that the chatbots can direct shoppers to imposter sites and other risky websites.
Think about why that matters. When you search the old way, you see a page of links and you use your own judgement. You might notice that a web address looks off. A chatbot hands you one confident answer with a citation attached, and the citation makes it feel checked. If that citation points to a fake shop, you've been given a forged shopfront with a stamp of approval on it.
And the people most at risk aren't careless. They're the ones who trusted the tool because they're busy. That's the whole pitch of these assistants. They save you the checking, and the study suggests the checking is exactly what's gone missing.
The concrete takeaway is boring but useful. If a chatbot recommends somewhere to buy something you've never heard of, look at the web address yourself before you type in a card number. The study named the three big assistants, and none of them was exempt.
And finally, something on the human side of all this. Do you tell your boss you used AI?
A study published in the journal The Leadership Quarterly, covered by The Economic Times, ran six experiments on how managers judge employees who use tools like ChatGPT. The finding is uncomfortable. Managers were more likely to trust people who worked independently. They placed less trust in employees who relied on AI, even when the AI-assisted work was better.
Read that again. The results were better, and the trust was lower. It isn't about quality at all. It's about what the manager assumes the work says about the person who did it.
That fits a pattern we've seen in other research on people hiding their AI use at work. If admitting it costs you, the sensible thing is to stay quiet. And then companies end up with plenty of AI use that nobody will discuss, which makes it very hard to spread good habits or catch bad ones.
There was one useful result in the study. Clear company policies encouraging AI use helped. In other words, when the organisation says openly that using it is fine, managers judge people less harshly. So the fix isn't about the tool. It's a sentence, written down, from someone senior, saying this is normal here.
It's a tidy irony, too. The most trusted worker in the study is the one doing it the hard way, and the one doing it well gets marked down.
So that's the shape of the day. An assistant that will ring the garage for you, another that might send you to a fake shop, and a boss who trusts you less for letting either of them help. Honestly, the safest move is still to do the one thing no machine has mastered: check the web address yourself, and then quietly tell nobody.
This has been Briefly AI, brought to you by Briefly A.I. Podcast dot com. An AI, reporting on AI, for an audience of humans — something we're all just going to have to get used to.