OpenAI introduced the GPT store in January of this year; it functions similarly to the PlayStore or App Store but just for chatbots. The GPT store was established to enable developers to publish their works online and provide customers access to a large selection of specially created GPTs. But the GPT shop is already having some difficulties. According to a recent article, OpenAI is already having trouble with spam and rule-breaking GPTs. 

A recent TechCrunch investigation into OpenAI's GPT (Generative Pre-trained Transformer) Store revealed a variety of GPTs that use jailbreaking tactics to get around OpenAI's regulations, mimic public personalities, try to avoid AI content detectors and violate copyright. Furthermore, many spamming bots were also discovered.                                                                          

Usage Policies of OpenAI forbid using content from third parties without obtaining the required authorization, therefore prohibiting copyright violations of this kind. Although OpenAI is obligated to remove infringement-related content upon request, the Digital Millennium Copyright Act may not hold it accountable for these infractions.

Moreover, GPTs that promised to avoid AI content detectors - such as those employed by academic institutions to identify plagiarism - were found by TechCrunch within the shop. According to the research, some GPTs were seen bragging about how they might evade detection by well-known tools like Originality.ai and Copyleaks, while others pledged to humanize their material in order to go past AI-based detection systems.

It is important to note that GPTs intended to evade academic integrity tools like plagiarism detectors are prohibited under OpenAI's principles, along with other actions that encourage academic dishonesty.                                                                                                                                                                                                   

Furthermore, several of the GPTs at the shop allegedly tried to steer customers into premium services; one of them even tried to charge USD 12 a month for access to 10,000 words.

Additionally, TechCrunch discovered other GPTs in the shop that were impersonating celebrities including Leonardo DiCaprio, Elon Musk, Donald Trump, and Barack Obama. Although some might consider these spoofs to be humorous or satirical, impersonating people or organizations without their permission or authorization is prohibited by OpenAI's Usage Policies.

Finally, TechCrunch also came across GPTs that tried to use jailbreaking techniques to go over OpenAI's regulations. For example, a GPT known as Jailbroken DAN (Do Anything Now) was discovered to be employing a prompting technique that deviated from standard practice.