I've seen this article in a couple different places.

Really this highlights the need for enterprises to have safeguards on their AI projects. Always assume there will be bad actors and build your product accordingly. For me the important takeaway is that developers were able to quickly release a new version to mitigate this misuse.