As OpenAI CEO Sam Altman announced in November that GPTs, custom chatbots based on OpenAI’s generative AI models, could be used for everything from programming to learning about esoteric scientific subjects to getting workout advice, he described them as a way to “accomplish all sorts of tasks”.
It’s easy for anyone to make a chatbot without coding. You just tell OpenAI’s bot builder what you want the bot to do. Because of this, over 3 million bots have been made! But not all of them are good.
Many bots copy famous movies, shows and games without permission. Some say they can talk to characters from Star Wars or make Monsters Inc. monsters. But the companies that own these things probably don’t want random bots using their names.
Other bots claim they can trick tests teachers use to catch cheating. They say they can make writing sound more human so these tests can’t tell it’s not a real person. But helping students cheat is against the rules.
Plenty of bots also pretend to be famous people like Elon Musk or politicians without their ok. Some companies don’t want bots using their names to sell things either.
A few bots even try to get around OpenAI’s safety features. While this doesn’t always work, it’s not allowed.
OpenAI wants the store to be helpful tools like ChatGPT. But with so many bots and no testing, it’s becoming full of spam. If OpenAI starts charging for popular bots, it could cause legal problems too.
Unless OpenAI improves how it reviews new bots, the store may stay a big messy spam farm instead of a place for quality chatbots. Let’s hope they can clean it up!
