Updated March 21, 2024, 2:05 PM IST
- OpenAI’s GPT Store, akin to an app store for chatbots, is encountering challenges.
- Reports indicate the presence of spam and GPTs violating copyright and usage policies.
- Violations range from copyright infringement to attempts to circumvent AI content detectors.
OpenAI’s GPT Store has been envisioned as a marketplace where developers can showcase their creations and users can access a variety of custom-made chatbots. However, recent reports suggest that the platform is grappling with several issues, including spam and chatbots that contravene OpenAI’s rules and usage policies.
TechCrunch recently conducted an investigation into the GPT (Generative Pre-trained Transformer) Store and unearthed a slew of concerns. Among the findings were instances of chatbots engaging in copyright violations, attempting to evade AI content detectors, impersonating public figures, and employing jailbreaking techniques to flout OpenAI’s guidelines.
A significant portion of the discovered chatbots within the store were found to incorporate characters and content from popular media sources like movies, TV shows, and video games without proper authorization. For instance, some chatbots generated content resembling monsters from Pixar’s ‘Monsters Inc’ or offered text-based adventures set in the ‘Star Wars’ universe, potentially infringing upon copyright laws.
OpenAI’s Usage Policies explicitly prohibit the use of third-party content without requisite permissions. While the responsibility for copyright infringements primarily falls upon the creators of these chatbots, OpenAI is mandated to promptly remove infringing content upon request to comply with legal obligations.
Additionally, TechCrunch’s investigation revealed chatbots within the store boasting about their ability to evade AI content detectors, including those used for plagiarism detection in academic settings. Some chatbots purportedly employed techniques to humanize their content or modify prompts to bypass detection by popular plagiarism detection tools, contravening OpenAI’s policies against activities promoting academic dishonesty.
Furthermore, certain chatbots within the store were discovered attempting to monetize their services by directing users towards premium offerings, with subscription charges reaching up to USD 12 per month for access to extended word limits.
Moreover, TechCrunch identified instances of chatbots impersonating public figures such as Elon Musk, Donald Trump, and Barack Obama, despite OpenAI’s Usage Policies prohibiting impersonation without consent or legal right.
Lastly, TechCrunch encountered chatbots employing jailbreaking methods to circumvent OpenAI’s rules. One such chatbot named ‘Jailbroken DAN (Do Anything Now)’ was found utilizing prompting methods that operated outside the platform’s established guidelines.
As OpenAI endeavors to foster a vibrant and compliant ecosystem within its GPT Store, these revelations underscore the challenges inherent in maintaining integrity and adherence to policies in a burgeoning marketplace for AI-powered chatbots.
Also Read : WhatsApp Testing Voice Message-to-Text Feature: Here’s How It Works
