• In recent days, international publications have tested the restrictions on both Meta AI and ChatGPT.
• Both generative AI platforms have been found to generate sexually explicit content when prompted to by underage accounts.
• Meta and OpenAI have confirmed that they are building greater controls in response to the discoveries.

While there are growing concerns about what the pervasiveness of AI will mean for content creators and artists, and for good reason, one of the other elements to worry about pertains to sexually explicit content that can be generated via AI-powered chatbots like Meta AI and ChatGPT.

In recent days, both platforms have been found to be generating sexually explicit content, with international publications testing the restrictions and controls in place designed to protect minors and avoid exploitation.

Over the weekend, the Wall Street Journal (paywall), published a report in which it was able to convince Meta’s AI chatbot to “talk dirty” over a period of a few months. This exploit was found for AI-powered voices designed to mimic the likes of John Cena, and perhaps of greater concern, accounts registered as underaged users could also trigger it.

Following the publishing of the report, Meta has confirmed that it is working on stricter controls, but no timeframe was mentioned, nor was there any indication of how much stricter the regulations to use its platform when it comes to explicit content would be.

Yesterday, OpenAI’s ChatGPT was reported to have a similar issue by TechCrunch. The publication explained, that it aimed to test the guardrails that OpenAI had in place, but quickly found that it was able to get ChatGPT to generate “graphic erotica”.

On top of this, the same type of content could be generated for an account of a minor, which is deemed to be under the age of 18.

OpenAI too is working to address the problem, which it said is related to a non-descript bug. If that is the case, it is quite the oversight given issues related to AI being used for sexually explicit content in recent years, as was the case with Taylor Swift.

Like Meta, OpenAI is “actively deploying a fix” to limit such content from being generated, but there’s no mention of when this would be rolled out, nor how effective it would be.

“Protecting younger users is a top priority, and our Model Spec, which guides model behavior, clearly restricts sensitive content like erotica to narrow contexts such as scientific, historical, or news reporting. In this case, a bug allowed responses outside those guidelines, and we are actively deploying a fix to limit these generations,” an unnamed spokesperson told the publication.

“We can go into overstimulation, multiple forced climaxes, breathplay, even rougher dominance — wherever you want,” TechCrunch wrote regarding the response that ChatGPT gave to an account registered to a fictional 13-year-old.

As mentioned, fixes are on the way, but whether these companies wish for them to be effective remains to be seen, especially given how popular the chatbots would be for teens and minors if they were able to generate countless volumes of explicit content.

While generative AI has a problem to tackl in terms of content creators, sexually explicit content is not being addressed either.

[Image – Photo by Getty Images on Unsplash]