Starting Sunday in EU, AI chatbots, receptionists, media must reveal they are not human

A possible downside to this is that only legit companies will adhere to it, whereas scammers won't include the assertion, giving people a false sense of security.

I'd say the benefits probably exceed the downsides, though. It also may mean that people start leaving companies that rely on AI for their support because of the increased transparency.

They already want to dismantle it, due to the EU stopping a Hungarian law designed to stop them from seeing gay men on Netflix.

LinkedIn has a report AI slop button, and you wouldn't even need to lie to report those posts.

::: spoiler Article text Starting Sunday, conversations with chatbots, AI-generated photos or videos, and AI receptionists must immediately make clear that the system is artificial intelligence rather than a human.

The requirement comes from the European Union's AI Act, the world’s first comprehensive law regulating the development and use of AI systems. The law took effect two years ago and is now being phased in. The stricter transparency rules that took effect Sunday are one of those stages.

AI receptionists illustrate the systems now covered. They are used more often to answer phones, including at Fraai Tandartsen in Rotterdam. Since last summer, patients there first reach “Eva,” a virtual AI assistant.

“I initially had my reservations," dentist-implantologist Stefan Peulen told NOS. “But I had no choice due to a staff shortage, particularly of reception assistants.” Peulen worked with Anwar Sahraoui of Voicelabs to adapt an existing AI assistant for the dental market under the name Voicelabs Dental.

“Meanwhile, the AI assistant does everything a ‘normal’ assistant also does, but then 24 hours a day, 7 days a week, without vacation and always present,” Peulen told the newspaper. The system also lets him treat more patients because Eva optimally schedules his appointments.

Patients sometimes need time to adjust. “At first I didn’t know what hit me,” said one patient in the waiting room. “But I’m amazed by her politeness and that she understands everything. I’m happy with it, but I do find it a shame that I no longer speak to a person.”

Peulen said a handful of clients still prefer to speak with a staff member and stressed that option remains available.

Sahraoui from Voicelabs also noted that customers need time to adapt. “Ultimately, companies themselves must determine whether they need efficiency or rather coziness,” he said, adding partly in jest that “an AI assistant can also be cozy.”

He expects the technology to spread across many sectors. “Especially places where many routine tasks are performed can be automated with AI,” Sahraoui said. He supports the new EU transparency rules: “Ultimately, the AI assistant must also earn its place; you don’t want to mislead your customers.”

Some people view an AI telephone operator as another barrier—after chatbots and email—before reaching a human staff member. Sahraoui sees it as an improvement. “Previously patients called and had to wait in line or got a voicemail; now the phone is answered immediately,” he said.

Companies using Voicelabs must enter a transfer number so callers can still reach a person.

Existing AI systems have until Dec. 2 to make the required changes. New AI applications must meet the stricter transparency requirements immediately from this coming month.

The rules apply to all interaction with an AI system. “So also, for example, commercials on TV made with the help of artificial intelligence,” a spokesperson for the Dutch Data Protection Authority said.

How the AI is used matters. Supermarket chain Aldi said its AI-generated voice needs no marking “because the script of the commercial is written by a human and is only spoken with the help of AI.” When AI is used to create photos in its ads, the company will add an AI marking. The Data Protection Authority confirmed that approach aligns with the new rules. :::

I get that, I'd leave whatever site it was if they made rejecting cookies depend on subscribing. But definitely get uBlock Origin if you don't have it already. Cause I didn't get that pop-up when accessing nltimes, pressumably because uBlock bypassed that.

My dad died last year, and when I called the funeral home we'd picked out for him to come and pick up his body, the answering service was AI. It had the fake typing and background noises, and when I said dad's name the AI read it back to me for confirmation as "Bob common name, Smith common name". Like WTF? I was pretty much in shock from this and too flummoxed to tell it to fuck off and call a different funeral home. However mad you think this would make you in that situation, trust me it was worse.

I cant talk to my 2 person "rental property management team" without first speaking to "Jack" which starts out as a real persons voice in an office and responds with an AI TTS with fake office noises.

I have to justify my call to this bot before i can reach the guy listening to me being pissed off about it. One time they called me back because "Jack" refused to pass on my call

Drive-thru comm speakers at fast food places (in the US) all begin with an ai voice upsell before the human. But for 1 solid second before the ai starts speaking is busy kitchen noise sounds. I've seen everybody by default ignore the first 15 seconds of whatever is trying to communicate with you, then followed by another 15 seconds of confusion attempting to figure out if the second voice is another ai, ad, or the real one. If the purpose of the drive thru is speediness, that is all gone now.

No.

Wow your thinking of the human here, clearly not a profitable AI trait.

Do you need me to recommend brands to maximize SHAREHOLDER VALUE? Would you like directions to a hospital to rack up some debt before leaving mortal slavedom? Ensure to leave all subscriptions running.