How Googlers cracked OpenAI’s ChatGPT with a single word

Training data was exposed. This could be bad. I’m not seeing this story picked up as the big story it appears to be?