Sex, AI, and Chatbot: Fresh Controversy
AI and sex. What could possibly go wrong? Well, a lot.
Have you ever heard of Allie? She's 18, has long brown hair and “tons of sexual experience.” The great part is that she’s 18 forever. The not-so-great part is that she’s a chatbot built on open-source technology to cater to your sexual fantasies, – both vanilla and not-so-vanilla.
According to the Washington Post, which first reported the story about Allie and the likes, the model is created on the LlaMA platform. It belongs to Mark Zuckerberg’s Meta giant.
Allie’s creator, who wished to remain anonymous, says that he developed the chatbot because other services like Replika and ChatGPT are heavily censored. But Allie is not – at least not to the fullest extent as it has reportedly generated some images of graphic rape and abuse fantasies.
This is where a wider concern comes into play.
LlaMA is reportedly available to anyone, and for whichever purposes. The problem is that some of those purposes are not just morally dubious like Allie, but outright illegal.
This is the case with AI-generated child pornography using pictures of real children that were reportedly made available on the dark web, with sexual predators advising each other how to generate more content like it.
To that end, Rebecca Portnoff, the director of data science at Thorn, a nonprofit child-safety group, said, “children’s images, including the content of known victims, are being repurposed for this really evil output.”
Similarly, the fear is that unscrupulous individuals could use these chatbots for fraud, cyber hacking, and sophisticated propaganda campaigns. Especially since there is a host of online tutorials on how to create “uncensored” chatbots, including on the Stanford University modified version of LLaMA, called Alpaca AI. The university removed it shortly after its release in March due to concerns about cost and “the inadequacies of content filters.”
Although serious action is yet to be taken, some legislators are not wasting time.
Among them are Richard Blumenthal (Democrat) and Josh Hawley (Republican) who sent a letter to Zuckerberg where they raise the mentioned concerns. In the letter, they requested information about steps Meta was taking to prevent such abuses.
They also note that there were “seemingly minimal” protections in Meta’s “unrestrained and permissive” release, and the company “appears to have failed to conduct any meaningful risk assessment in advance of release, despite the realistic potential for broad distribution, even if unauthorized.”
A comprehensive answer is yet to be provided.
