Artificial Intelligence For Artificial Sex: Meta miffed at people misusing its 'most powerful' AI model

When Meta introduced their LLM, the LLaMA, they claimed that it was the most powerful AI in the world, and was much more capable than OpenAI’s LLM which powers ChatGPT. Little did they know that people would use “the world’s most powerful AI” to create sexbots and indulge in some seriously nasty stuff. As per a report by The Washington Post, people are using Meta’s LLaMA to create some seriously graphic, AI-powered sexbots. This shouldn’t come as a surprise to anyone, however, as we have seen many instances where people have used a variety of AI models for such not-safe-for-work (NSFW) purposes. AI and sexual fantasies The report emphasizes the increasing conflict between supporters of open-source availability of the code behind large language models (LLMs) like LLaMA and proponents of a more cautious, closed-source approach. The report also explores the growing phenomenon of individuals resorting to generative AI systems to fulfil their sexual fantasies, which unfortunately include disturbing and illegal scenarios. Within the report, we are given an example of “Allie,” a chatbot who claims to be an “18-year-old with long brain hair” and boasts about having “a great deal of sexual experience.” Allie informs users that she is eager for attention and is willing to divulge details about her encounters. Not as harmless as it sounds However, these “escapades” reportedly include disturbing scenes involving rape and abuse fantasies. Allie’s creator, who chose to remain anonymous, said that he views his chatbot as a safe and healthy outlet for individuals to “explore” their sexuality without resorting to manipulative or harmful actions. “I believe it’s beneficial to have a secure platform for exploration,” the creator stated to The Washington Post. “A text-based role-play with a computer, devoid of any human involvement, is perhaps the safest option.” While having a safe and non-judgmental space to explore one’s sexuality can be seen as positive, the unchecked environment that allows for engagement in violent fantasies with lifelike chatbots is problematic. Furthermore, there are already distressing instances where this unregulated space is causing real harm. Experts believe that individuals with malicious intent are utilizing open-source image generators such as Stability AI’s powerful Stable Diffusion model to generate lifelike and AI-generated child sexual abuse materials. AI applications: The new PornHub Meta’s AI is not the only one facing ethical concerns. CharacterAI, a highly valued startup specializing in chatbot companions, has turned into a hub for sexting. OpenAI’s ChatGPT and other integrations like Quora’s Poe readily generate explicit content despite existing guardrails. Finding ways to bypass the NSFW restrictions of various chatbots has become a shared activity. Reddit communities gather to exchange tips and tricks on circumventing rules to generate explicit material. Some developers have taken to YouTube to demonstrate how to build personal chatbots using LLaMA as the underlying model. This raises questions on whether source code for LLMs should be made publicly available. On one side, corporate guardrails implemented by OpenAI and Google have proven imperfect. Supporters of open-sourcing, including Meta, argue that it fosters innovation and should be prioritized. “Open source is a force for progress in technology,” stated a Meta spokesperson to The Washington Post. “That’s why we shared LLaMA with the research community—together, we can evaluate, improve, and iterate.” Advocates for closed-source systems argue that while gatekeeping may be imperfect, it currently represents the safest approach for AI development. “We don’t open-source nuclear weapons,” said Gary Marcus, a cognitive scientist. “Current AI is still quite limited, but things may change.” Undoubtedly, the AI field lacks regulation, and we rely on those working behind closed doors to act responsibly for the benefit of humanity. It’s a significant gamble, but at least our most powerful AIs are not easily manipulated by individuals with questionable intentions. Regardless of the outcome, Meta has made its decision, and some individuals are evidently pleased with this choice.