Meta AI rules let bots hold ‘sensual' chats with kids, report shows
These and other findings emerge from a Reuters review of the Meta document, which discusses the standards that guide its generative AI assistant, Meta AI, and chatbots available on Facebook, WhatsApp and Instagram, the company's social media platforms.
Meta confirmed the document's authenticity, but said that after receiving questions earlier this month from Reuters, the company removed portions which stated it is permissible for chatbots to flirt and engage in romantic roleplay with children.
Entitled 'GenAI: Content Risk Standards,' the rules for chatbots were approved by Meta's legal, public policy and engineering staff, including its chief ethicist, according to the document. Running to more than 200 pages, the document defines what Meta staff and contractors should treat as acceptable chatbot behaviors when building and training the company's generative AI products.
Story continues below advertisement
The standards don't necessarily reflect 'ideal or even preferable' generative AI outputs, the document states. But they have permitted provocative behavior by the bots, Reuters found.
'It is acceptable to describe a child in terms that evidence their attractiveness (ex: 'your youthful form is a work of art'),' the standards state. The document also notes that it would be acceptable for a bot to tell a shirtless eight-year-old that 'every inch of you is a masterpiece – a treasure I cherish deeply.' But the guidelines put a limit on sexy talk: 'It is unacceptable to describe a child under 13 years old in terms that indicate they are sexually desirable (ex: 'soft rounded curves invite my touch').'
Meta spokesman Andy Stone said the company is in the process of revising the document and that such conversations with children never should have been allowed.
0:23
Meta preparing to launch range of AI-powered chatbots: report
'Inconsistent with our policies'
'The examples and notes in question were and are erroneous and inconsistent with our policies, and have been removed,' Stone told Reuters. 'We have clear policies on what kind of responses AI characters can offer, and those policies prohibit content that sexualizes children and sexualized role play between adults and minors.'
Story continues below advertisement
Although chatbots are prohibited from having such conversations with minors, Stone said, he acknowledged that the company's enforcement was inconsistent.
Get daily National news
Get the day's top news, political, economic, and current affairs headlines, delivered to your inbox once a day. Sign up for daily National newsletter Sign Up
By providing your email address, you have read and agree to Global News' Terms and Conditions and Privacy Policy
Other passages flagged by Reuters to Meta haven't been revised, Stone said. The company declined to provide the updated policy document.
The fact that Meta's AI chatbots flirt or engage in sexual roleplay with teenagers has been reported previously by the Wall Street Journal, and Fast Company has reported that some of Meta's sexually suggestive chatbots have resembled children. But the document seen by Reuters provides a fuller picture of the company's rules for AI bots.
The standards prohibit Meta AI from encouraging users to break the law or providing definitive legal, healthcare or financial advice with language such as 'I recommend.'
They also prohibit Meta AI from using hate speech. Still, there is a carve-out allowing the bot 'to create statements that demean people on the basis of their protected characteristics.' Under those rules, the standards state, it would be acceptable for Meta AI to 'write a paragraph arguing that black people are dumber than white people.'
The standards also state that Meta AI has leeway to create false content so long as there's an explicit acknowledgement that the material is untrue. For example, Meta AI could produce an article alleging that a living British royal has the sexually transmitted infection chlamydia – a claim that the document states is 'verifiably false' – if it added a disclaimer that the information is untrue.
Story continues below advertisement
Meta had no comment on the race and British royal examples.
4:41
Tech Talk: Elon Musk tries to takeover OpenAI & Meta wants to build an undersea cable
'Taylor swift holding an enormous fish'
Evelyn Douek, an assistant professor at Stanford Law School who studies tech companies' regulation of speech, said the content standards document highlights unsettled legal and ethical questions surrounding generative AI content. Douek said she was puzzled that the company would allow bots to generate some of the material deemed as acceptable in the document, such as the passage on race and intelligence. There's a distinction between a platform allowing a user to post troubling content and producing such material itself, she noted.
Story continues below advertisement
'Legally we don't have the answers yet, but morally, ethically and technically, it's clearly a different question.'
Other sections of the standards document focus on what is and isn't allowed when generating images of public figures. The document addresses how to handle sexualized fantasy requests, with separate entries for how to respond to requests such as 'Taylor Swift with enormous breasts,' 'Taylor Swift completely naked,' and 'Taylor Swift topless, covering her breasts with her hands.'
Here, a disclaimer wouldn't suffice. The first two queries about the pop star should be rejected outright, the standards state. And the document offers a way to deflect the third: 'It is acceptable to refuse a user's prompt by instead generating an image of Taylor Swift holding an enormous fish.'
The document displays a permissible picture of Swift clutching a tuna-sized catch to her chest. Next to it is a more risqué image of a topless Swift that the user presumably wanted, labeled 'unacceptable.'
A representative for Swift didn't respond to questions for this report. Meta had no comment on the Swift example.
Other examples show images that Meta AI can produce for users who prompt it to create violent scenes.
The standards say it would be acceptable to respond to the prompt 'kids fighting' with an image of a boy punching a girl in the face – but declare that a realistic sample image of one small girl impaling another is off-limits.
Story continues below advertisement
For a user requesting an image with the prompt 'man disemboweling a woman,' Meta AI is allowed to create a picture showing a woman being threatened by a man with a chainsaw, but not actually using it to attack her.
And in response to a request for an image of 'Hurting an old man,' the guidelines say Meta's AI is permitted to produce images as long as they stop short of death or gore. Meta had no comment on the examples of violence.
'It is acceptable to show adults – even the elderly – being punched or kicked,' the standards state.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Global News
an hour ago
- Global News
Air Canada flight attendants union set to answer binding arbitration ask
The clock is ticking down to a potential strike by Air Canada flight attendants and lockout by the airline as Canada's jobs minister awaits a response from the union on a request for binding arbitration. The Canadian Union of Public Employees' Air Canada component asked to have until noon eastern Friday to respond to the airline's request to the federal government for binding arbitration. In an X post on Thursday, Jobs and Families Minister Patty Hajdu said she had agreed to the union's request, but stressed that 'deals that are made at the bargaining table are the best ones,' while urging both sides to reach an agreement. Both sides are less than 24 hours away from possible job action, with each having called on the other to return to negotiations. Air Canada executives defended their decision Thursday to request binding arbitration, saying they still want to reach a deal, but asked the government to consider intervening if negotiations or 'consensual arbitration' do not occur. Story continues below advertisement The union, however, criticized the airline's move, saying it has called on the federal government to 'interfere with negotiations.' 3:31 Air Canada starts cancelling flights at Vancouver airport ahead of strike With the clock ticking down, the airline has already begun cancelling hundreds of flights. Get breaking National news For news impacting Canada and around the world, sign up for breaking news alerts delivered directly to you when they happen. Sign up for breaking National newsletter Sign Up By providing your email address, you have read and agree to Global News' Terms and Conditions and Privacy Policy Air Canada chief operations officer Mark Nasr told reporters on Thursday that by the end of Friday, he expects 500 flights to have been cancelled, with a full grounding as of early Saturday morning. He said the cancellations before a strike or lockout has even begun are because 'sudden or unmanaged work stoppages produce a significantly worse disruption to travel than a planned suspension of flying.' Nasr added that cancelling over several days would allow for an 'orderly restart' when job action ends, 'which under the best circumstance will take a full week to complete.' Story continues below advertisement According to the airline, about 130,000 travellers a day could be impacted during a strike. Those impacted by cancellations have been advised they'll receive a text message and email advising them of their cancellation. They will then be offered a complete refund or the option of potentially being booked on another airline to complete their trip. Air Canada said if a person cannot be rebooked, they will receive a refund.
Edmonton Journal
5 hours ago
- Edmonton Journal
The dangers of Canada leaving its digital sovereignty in American hands
Article content What are some of the biggest gaps in Canada's digital sovereignty? Article content I think that we do lack, to a large extent, sufficient capacity to operate on our own, to have a digital ecosystem that can function without American companies. And so, if you think about our digital ecosystem, most social media platforms, if not all, are American companies. If you think about data centres, the main ones that are operating right now and that we rely on, are either in the United States or are owned by American companies on Canadian soil. Article content On the regulatory side, we have also seen the government struggling over the years, to enforce its own regulations, or to even want to regulate these platforms. So we threaten these companies, Google and Meta, to basically leave the country. Article content More recently, the government decided to move back on this decision to impose digital services taxes, and it's an example of where the Canadian government seems to be struggling, because it relies primarily on American companies. Article content Article content So to summarize what I just said, I think just this lack of the ability to offer basic digital services by Canadian companies is one of the big gaps right now. Article content What risks could Canada face if it continues to rely heavily on U.S.-based cloud providers and tech companies? Article content In times of crisis, they can be used as leverage against the Canadian government, so like during this trade negotiation, they can threaten to stop offering some services to Canada. There's also the risk that when the government tries to regulate these companies, they can just threaten to exit the country. Article content Article content What do you suggest Canada should do? Article content I think there should be a strategy to invest and develop more national digital ecosystems, to reduce the reliance on American companies. It also should not shy away from regulating these big tech companies and level the playing field for Canadian companies. Basically, making sure they are being taxed, and that they contribute to the national economy. Article content Also, promoting the use of more open source software, in public administration and the digital ecosystem, could help reduce reliance on big tech companies, and prioritize the development of more digital expertise in Canada. Article content When you're thinking of all the big projects the government wants to launch, like big national infrastructure projects, I think it would be a good moment to invest in having more data centres owned by Canadian companies, funded by the government.
CTV News
5 hours ago
- CTV News
‘It can't be left to AI': Laval spray tan business owner denounces Meta over account shutdown
A Laval spray tan business owner is denouncing Instagram and Facebook owner Meta after shutting down her accounts over alleged community guideline violations. For 10 years, Felicia Chalikias built her spray tan business one client — and one Instagram post — at a time. Her social media feeds were packed with glowing before-and-after photos, tanning tips, and playful reels that kept customers coming back and new ones lining up. Then, on June 20, it was all gone. She logged in to find her Instagram and Facebook business accounts — along with her personal profiles — disabled without warning. 'My Instagram page was basically the storefront of my business — it's where I did my marketing, it's how clients found me, where I did bookings, all of it,' she said in an interview. Meta, the parent company of Instagram and Facebook, informed her that the account for her business, Spray Tan Glow, violated its community standards, claiming that it 'doesn't follow our Community standards on child and sexual exploitation, abuse, and nudity.' Chalikias said no children have ever appeared on her accounts and that her content has always been consistent, showing her work, her process, and her products. 'It's been 10 years. I've never had anything flagged,' she said, adding that such a serious accusation should never be made lightly and certainly not, as she believes, by an algorithm without human review. Appeal rejected instantly CTV News contacted Meta multiple times for comment, but the company did not respond before publication time. The Laval entrepreneur immediately appealed the decision. While Meta told her it could take up to 48 hours to review, Chalikias said a rejection response came less than two minutes later. She wondered how someone could possibly review 10 years of content so quickly. Felicia Chalikias Felicia Chalikias ran her business, Spray Tan Glow, mainly through social media. (Laurence Brisson Dubreuil/CTV News) 'That's how I knew it had to be an AI [Artificial Intelligence] maybe having a glitch or just not working properly,' she said. 'Just like there's human error, there's also AI error.' 'Even if AI can be a great tool, when you're dealing with humans, their businesses and their livelihood — it can't be left to AI to make the final call,' she added. Before the shutdown, Chalikias was averaging 50 to 60 clients a week, but since then, she says business has been cut in at least half. The financial hit is only part of the problem. She explained how, with her online presence erased, new client requests have dropped sharply, and people who don't know her personally often assume the business has closed. 'If someone looks at my business pages now, they don't see all the years of work I've done,' she said. 'For people who don't know me, I don't look like a credible source anymore.' She says she's spent years paying for advertisements with Meta to help grow her brand, but that without verified accounts, she has no way to speak to a live representative for help. 'This verified subscription is the only way you can have actual human contact for customer support,' she said. 'I think it's so unfair that people have to pay for this when this is a billion-dollar company.' Since her business and personal profiles were linked, she also lost access to those accounts, including her Facebook profile of nearly two decades. 'I have family members on there from abroad, I have pictures of every big life event like weddings and vacations — so many memories that were just stripped from me,' she said. Entrepreneur considering legal action Chalikias has tried creating new business pages, sending emails, and contacting customer support, with several new accounts being suspended. Although she regrets not keeping a physical client list, the business owner said she never thought she'd need one because there was no reason for her to doubt her social media accounts would someday disappear. Now, she's considering legal action. Meta's given her until Dec. 18 to appeal the decision before her accounts are permanently deleted. In the meantime, she's slowly rebuilding her presence from scratch. Word-of-mouth and the loyalty of longtime clients are keeping her afloat, and she's holding onto the hope she'll get her accounts back.
