cross-posted from: https://feddit.online/c/tech@piefed.world/p/2037684/how-far-will-ai-assistants-go-for-a-refund-they-cant-get-spamming-the-government-using
This is essentially the “paper-clip factory” alignment problem.
It’s good to remember that this same kind of behavior was seen in the hugging face hack and other similar hacks.
Prompt agent to accomplish task -> agent “believes” it cannot accomplish it (this does not need to actually be true) -> agents context window begins fill -> agent’s initial prompt is condensed over time until it’s own training and guardrails fade into obscurity -> agent convinces a 13 year old in Luxemburg to jump off a bridge as a part of a large and very elaborate plan to get the developers of an npm package you’re using for your project to include a feature the agent is convinced it needs to complete its task
What was the point? All I gathered from reading this is that each AI agent has a different way of lying.
…I think that’s an excellent point to take away from this article. Tech companies WANT this future. They want agent to agent interactions that you are too busy or lazy to keep track of.
We should all be aware of what we could potentially be subjecting others to by handing our affairs over to these half baked bots.
Can’t wait for the first harassment case filed against someone’s personal agent.
Should be filed against that someone directly, actually. If you ghost ride the wheel and run over a toddler, they don’t charge the automatic transmission for creeping at 8 km/h, they charge you.
I’ve had great success using agents to play multiple accounts to get cheap shit of Facebook marketplace
This is why we have a harder and harder time using any online marketplace. We want our stuff to go to an actual human being and don’t charge very much for it, but the hassle of being bombarded with super lowball bullshit offers and from accounts that clearly aren’t real people has soured us.
Now we just stick most of our shit out for bulk pickup a couple days early and hope someone gets a chance to pick up anything useful.
What exactly are you looking for on FB marketplace that’s worth the time and effort of setting up and running agents to monitor it? Why do you need multiple accounts?




