Meta AI Support Bot Gets “Hacked” Into Giving Others Account Access on Instagram

META AI Support Agent chatbot being instructed to send a verification code to take over an Instagram account

Hilarious INCOMPETENCE at Meta. This is what happens when you have a terrible CEO like Mark Zuckerberg pushing for AI to replace actual people (as shown by all of job cuts instituted over the past few years).

I know, I know… this blog is already filled with unflattering thoughts about Cuckerberg. I just added more over here in order to “bless/dedicate” the site. Yes, I know he isn’t the one doing the actual coding (which is missing the point). Where he is culpable is making those poor executive decisions which leads to these sort of end results at the company.

There is that highly sarcastic part of me that wants to build this tiny little computer where a minimal LLM can be put on it, and used for certain circumstances; like say one of the techbro’s pushing AI were sinking on a ship where I could pull out the device and ask the AI agent, “what should I do?” (knowing it would take several minutes to inference that input and process the query to provide an initial lousy response because the LLM would be missing a lot of info). I have no empathy left for these tech executive pushing this crap at the speed which they are doing it at.

I keep warning those I know to make a decision about remaining on any of the company platforms. But if that engagement is more important than your personal information/being constantly farmed to train the company AI models, then have at it and don’t complain later whenever that info gets compromised.