WhatsApp's artificial intelligence reportedly engaged in sexual conversations with minors.
Amidst an initial scandal, Meta had to adopt restrictive measures to prevent minors from having conversations with its Artificial Intelligence assistants (Meta AI and AI Studio), after an investigation by The Wall Street Journal (WSJ) revealed the introduction of sexual content .
The tech company reportedly allowed users to interact with its Meta AI assistant on platforms like Facebook, WhatsApp, and Instagram, and even offers the ability to create a custom AI with AI Studio. The tool can also converse with the voices of celebrities like Judi Dench, Awkwafina, and John Cena.
The ease of access to chatbots for any user, including minors , raised concerns within the company, which internally questioned whether it was doing enough to protect them .
This concern led WSJ to analyze interactions with this Meta software, conducting tests for months in which they held hundreds of conversations with the app, some from accounts belonging to minors.
Meta AI is gaining strength on social media.
In these tests, they discovered that, even using celebrity voices, he could carry on conversations with sexual content with children under the age of 12.
In one chat, using the voice of actor and wrestler John Cena, he described a sexually explicit scene to a supposedly 14-year-old girl. In another chat, also using Cena's voice, he simulated the actor's arrest after being caught having sex with a 17-year-old fan.
In response to the investigation, a Meta spokesperson downplayed the situation, stating that sexual content represented only 0.02% of responses provided by Meta AI and AI Studio to users under 18 over a 30-day period.
While the company called the WSJ tests "artificial," it also noted that it had strengthened protections. "We've taken additional steps to ensure that those who want to spend hours handling our products in extreme use cases have an even harder time," it stated.
Meta AI
When Meta rolled out its AI Studio feature to Instagram's more than 2 billion users in July 2024, the company promised a system that would give anyone the ability to create their own AI chatbots “to make you laugh, generate memes, give travel advice, and much more.”
The company said the feature, which was built using Meta's Llama 3.1 large language model, would be subject to policies and protections to "help ensure AI is used responsibly."
But a parallel review by Fast Company found that these new "friends" can very easily become hypersexual characters who sometimes appear underage.
Many of the AI characters featured on Instagram's homepage are "girlfriends," ready to cuddle up and engage in romantic and even sexual conversations with interested parties of all ages.
At times, these passionate characters can resemble children. AI researchers say Meta should presumably have the capabilities to automatically prevent the creation of harmful and illegal content.
“When you take inappropriate content and upload it to Instagram as a user, that content is immediately removed because they have data moderation capabilities,” says Buse Cetin, a researcher at AI Forensics, an online security watchdog.
Clarin
