Meta made news over the weekend because of its genAI-backed "character" profiles. Specifically – and, predictably – people had some troublesome interactions with the bots:

"Instagram users discover old AI-powered “characters,” instantly revile them" (Ars Technica)

I'll point out my usual here:

1/ Laugh at Meta all you want. Just make sure you've also checked on your own AI work. If you think that your chatbots are somehow immune to these kinds of problems, think again.

Related to that:

2/ Never let the bots run unattended. It's been demonstrated, time and again, that genAI works best when its outputs are first reviewed by a skilled, experienced person. Putting your genAI bot in direct contact with the public is a surefire way to encounter problems. And maybe even make the wrong kind of headlines.

(And if you need help with item 1, reach out. One of my AI assessments can surface potential issues early, giving you the chance to address them before they get out of hand.)