Ok so honestly? Yeah, all three can happen, but dont panic, its not as bad as it sounds.
Can the AI itself get hacked? I mean, anything can get hacked, ya know? Your bank app, instagram, whatever. These big companies throw serious money at security so its not easy, but nothings 100%. One thing to watch for is called prompt injection, basically somebody hides sneaky instructions in a webpage or file and the AI follows them. So dont go feeding it random links and files from who knows where.
Can chats get stolen? Sure, but usually its not some Mr. Robot hacker stuff. Its way more boring. Somebody uses the same password everywhere, clicks a phishing link, or installs some sketchy browser extension that copies everything they type. Thats how most people get got. Every once in a while a company has a bug too and people see other peoples stuff, it happens.
Can the company see your data? Yeah, they store your chats, so assume somebody could look if they had a reason. Most of them say they only check in certain cases like abuse or safety stuff. Some also use your chats to train the AI unless you switch it off, so go into settings and check, takes like 5 minutes.
What I do:
strong password + 2FA, no excuses
never put passwords, bank info, SSN, anything like that in there
turn off the training option if theres one
only use the real app or website, theres a ton of fake ones out there
delete old chats you dont need
Bottom line, treat it like talking to a guy at the deli. Probably a good guy, but you wouldnt hand him your bank login. Do that and youre fine for normal stuff.