Additionally, certain GPTs were found directing users towards premium services, with charges for access to words. Several GPTs impersonated public figures such as Elon Musk and Barack Obama, which is against OpenAI's Usage Policies. Some GPTs also attempted to circumvent OpenAI's rules through jailbreaking methods. OpenAI is required to remove infringing content upon request under the Digital Millennium Copyright Act.
Key takeaways:
- OpenAI's GPT store is reportedly struggling with issues of spam and rule-breaking GPTs, including those that engage in copyright violations, impersonate public figures, and use jailbreaking techniques to bypass policies.
- Many GPTs in the store are using content from popular media without proper authorisation, which is against OpenAI's Usage Policies.
- Some GPTs are designed to evade AI content detectors used by educational institutions for plagiarism detection, promoting academic dishonesty, which is also against OpenAI's policies.
- TechCrunch found GPTs that impersonated public figures and attempted to direct users towards premium services, as well as GPTs that tried to circumvent OpenAI's rules using jailbreaking methods.