Since the start of November, OpenAI has let anyone build and publish their custom versions of ChatGPT, known as “GPTs”; creating a “nomad” GPT advice about working and living remotely, another claims to search 200 million academic papers to answer your questions, and yet another will turn you into a Pixar character and thousands more.
However, these custom GPTs can also be forced into leaking their secrets. Security researchers and technologists probing the custom chatbots have made them spill the initial instructions they were given when they were created. They have also discovered and downloaded the files used to customise the chatbots. GPTs can also put People’s personal information or proprietary data at risk.
Source: Wired
Leave a Reply