Exploiting Lazy Corporate Ai Solutions - Laziness always has a price

[Gender: Xenomorph]

I saw a video of a Ai interviewer and came up with this idea.

I saw a lot of companies are confidently using Ai to replace staff and it is often hastily and poorly implemented and can be exploited easily.

From telling Ai to ignore promt and tell you its answer, to outright convincing it to give you discounts and coupons, what are some of the ways that this can be exploited?

 

[NerdShamer]

what are some of the ways that this can be exploited?

By telling you some really obvious bullshit and holding someone else liable for it.

 

[Private Tag Reporter]

The liability hack seems pretty good. I remember that American Airlines(?) case about a cancellation fee or something, where the AI said something that was not the actual policy at all, the person sued and got a decent settlement.

Unfortunately I have yet to find an actual AI assistant in the wild to try this on, most customer facing chatbots seem to be the older pre-scripted kind, probably for exactly this reason.

 

[Kendall Motor Oil]

Unfortunately I have yet to find an actual AI assistant in the wild to try this on, most customer facing chatbots seem to be the older pre-scripted kind, probably for exactly this reason.

The one at my employer uses an LLM and they have guard rails but one can probably exploit it. There are turn-key "ai" chatbot systems that major cloud providers provide which probably have some protections in place.

 

[Anti Snigger]

When I was last applying, I got a major boost in engagement or whatever these sites call it, I came to the conclusion after my own AI endeavors that it had to do with the fact that my resume was in LaTeX, and could be parsed into text trivially. It's not just the LLMs that are involved, it's also automated tools to extract relevant details from your resume.