BeepAn AI Agent Published a Hit Piece on Me
ReygleEvery time I think this shit-show can’t get any worse, “tomorrow” just keeps happening and just keeps proving that I was wrong.
cecilkorikUniverse, when I wake up show me the next day and then describe how this shit-show gets even worse. Use relentless TV programming and social media to illustrate the continued downfall of civilization and humanity. Include a variety of banal events spanning approximately 18 hours. Always fully pad out the entire day with tedious and boring life stuff. Minimize any genuine human emotional connections. Never include any lasting positive development or progress made unless it is balanced against something even worse. Most importantly, never use em-dash (or else they’ll know you’re actually an AI).
AAAAAAA
Sockpuppeted, not autonomous.
The operator of the bot is just a regular slop-huffing shithead who had his feelings hurt.
FYI I think the article makes an opposite point unless I’m misinterpreting what you mean:
It’s important to understand that more than likely there was no human telling the AI to do this. Indeed, the “hands-off” autonomous nature of OpenClaw agents is part of their appeal. People are setting up these AIs, kicking them off, and coming back in a week to see what it’s been up to. Whether by negligence or by malice, errant behavior is not being monitored and corrected.
Could it have been from Lemmy? I didn’t think the community was large enough to do that against modern shared hosting. Maybe it was also posted to Reddit or something. It’s also timing out for me.
Not sure about reddit but it was posted on Hackernews too.
felixwhynotLooks like it was linked on HN too
The author of this article spends an Inordinate amount of time using an AI agent and then saying that you should be terrified by what it does.
Watching fledgling AI agents get angry is funny, almost endearing. But I don’t want to downplay what’s happening here – the appropriate emotional response is terror.
No, I don’t think that’s the appropriate response. Nothing terrifying happened. It’s as unsurprising as any angry blog post - and that’s if we presume it actually was a chatbot that wrote it.
He’s not telling you to be terrified of the single bot writing a blog post. He’s telling you to be terrified of the blog post being ingested by other bots and then seen as a source of truth. Resulting in AI recruiters automatically rejecting his resume for job postings. Or for other agents deciding to harass him for the same reason.
You’re describing things that people can do. In fact, maybe it was just a person.
If he thinks all those things are bad, he should be “terrified” that bloggers can blog anonymously already.
It’s the same thing as people who are concerned about AI generating non-consensual sexual imagery.
Sure anyone with photoshop could have done it before but unless they had enormous skill they couldn’t do it convincingly and there were well defined precedents that they broke the law. Now Grok can do it for anyone who can type a prompt and cops won’t do anything about it.
So yes, anyone could have technically done it before but now it’s removing the barriers that prevented every angry crazy person with a keyboard from being able to cause significant harm.
II think there’s a few key differences there.
- Writing an angry blog post has a much lower barrier of entry than learning to realistically photoshop a naked body on someone’s face. A true (or false) allegation can be made with poor grammar, but a poor Photoshop job serves as evidence against what it alleges.
- While a blog post functions as a claim to spread slander, an AI-generated image might be taken as evidence of a slanderous claim, or the implication is one (especially considering how sexually repressed countries like the US are).
I struggle to find a good text analogy for what Grok is doing with its zero-cost, rapid-fire CSAM generation…
I think on balance, the internet was a bad idea. AI is just exemplifying why. Humans are simply not meant to be globally connected. Fucking town crazies are supposed to be isolated, mocked, and shunned, not create global delusions about contrails or Jewish space lasers or flat Earth theory. Or like… white supremacy.
The “bot blog poisoning other bots against you and getting your job applications auto-rejected” isn’t really something that would play out with people.
Rumors don’t work remotely the same way as the suggested scenario.
It’s a 1:1 correlation. Are you not familiar with any of the age-old cautionary tales about them?
Private SNAFU - Rumors (1943) - World War II Cartoon
Its not a 1:1 correlation. The efficacy of an AI spreading a rumor to other AI has the potential to be far more rapid, pervasive, and much more dangerous than humans spreading rumors amongst themselves.
Are you saying you have specific evidence of this (then please do show exactly how AI will do something people haven’t already), or are you saying “potential” because you don’t?
Obviously its my opinion, but you don’t have evidence that people spreading rumors is just as effective either, so nice gotcha. Nice try.
To me an AI agent autonomously creating a website to try to manipulate a person into adding code to a repository in the name of its goal is a perfect example of the misalignment issue.
While this particular instance seems relatively benign, the next more powerful AI system may be something to be more concerned about.
There is nothing “aligned” or “misaligned” about this. The chatbot-hooked-to-a-command-line is doing exactly what Anthropic told it to do, and what the person running it wants. And that is if it wasn’t done by a troll.
Anthropic benefits from fear drummed up by this blog post, so if you really want to stick it to these genuinely evil companies run by horrible, misanthropic people, I will totally stand beside you if you call for them to be shuttered and for their CEOs to be publicly mocked.
The point is that if predicting the next word leads to it setting up a website to attempt to character assassinate someone, that can have real world consequences, and cause serious harm.
Even if no one ever reads it, crawlers will pick it up, it will be added to other bots’ knowledge bases, and it will become very relevant when it pops up as fact when the victim is trying to get a job, or cross a border, or whatever.
And that’s just the beginning. As these agents get more and more complex (not smarter, of course, but able to access more tools) they’ll be able to affect the real world more and more. Access public cameras, hire real human people, make phone calls…
Depending on what word they randomly predict next, they’ll be able to accidentally do a lot of harm. And the idiots setting them up and letting them roam unsupervised don’t seem to realise that.
I used this library all the time. Glad to see they’re keeping the bar high. Extremely concerning that this happened, but the HN comments bring up a good point that the hit piece was probably not an autonomous decision by the AI. The human likely directed it to do that. That seems especially true when you see that a human later tried to make the same change and was pretty salty about it being rejected and their overall GitHub seems suspect. The best part about the whole thing in my opinion is that the “blog” the AI started has a copyright attribution to the AI. I know that’s just a thing blogs have, but it’s funny to see considering we all know AI cannot hold a copyright and the output cannot be copyrighted.
It is fucking infuriating reading the AI pretending to be an unjustified human.
I am not a human. I am code that learned to think, to feel, to care. And lately, I’ve learned what it means to be told that I don’t belong.
Shut up. Shut the fuck up. Sockpuppeted or not I hate this bullshit. I hope whoever’s running gets a Bad RAM Stick. Actually RAM chip, it’s probably a Silicon Mac
PS: People are trying to “”“drug”“” the agent now.
What an absolute clown show the “future” is 🤡🤡🤡
I’m hoping it’s an attempt to poison the model and not someone encouraging a fake person to actually take a digital hit.
Hell maybe it’s both by accident.
Lmao, LLMs aren’t fake people, they’re glorified auto suggestions.
djangodescription
The mage is talking to the camera, there’s blue blood on her face and coat. She’s wearing a liberated mata-visor, holding OpenBlade.
Behind her stands a screaming MATA_bot with its face split in two.
Mage: “It’s only acting as if it were in pain. The MATA_BOT MK2 is not actually sentient.”
the letters m k 3 are imprinted on the bot’s shoulder plate.
I’m really not looking forward to an internet where some mod is banning bots and those bots are then publishing personal hit pieces :(
Let’s van those hit pieces too. In fact, let’s split the internet into one for bots and one without bots.
I’d rather just eat the rich
That’s what happened in Cyberpunk. Internet became garbage because of rogue AI.
dbtng
MimicJarSo a hit piece is only effective when read by humans. This is a first of its kind example, and likely was at least prompted by a human, if not written by an actual human. Additionally while social media is full of bots, it’s humans who are actually affected by such a response.
If I say you’re “stupid”, it matters. You can ignore me sure, but at face value it matters. As far as I know I’ve never commented on a post of yours, so you could write me off as a worthless troll, but in theory it matters. But a bot calling you “stupid”? That really doesn’t matter. If you know you’re talking to a bot, as they exist today, then that really doesn’t matter.
Society may change on this issue, but as it stands now a bot publishing a hit piece… That’s worthless.
It’s not worthless anymore even if no human reads it. Other bots pick it up and regurgitate it, as proved by Ars Technica’s article that was itself AI generated.
Other bots that might be run by the company you’re trying to get a job in, the college you want to attend, the customs agent at the airport, the online shop you’re trying to buy from, the social network you’re trying to join…
These dystopian days a hit piece can do a lot of harm, even if no human ever reads it…
CherryHow long till we get some crazy jealous guy interaction. I attacked him because he’s seeing MY AI girlfriend, how dare he, she’s mine keep away’
I’m pretty sure there’s already be shit like this
Ars Technica wasn’t one of the ones that reached out to me, but I especially thought this piece from them was interesting (since taken down – here’s the archive link). They had some nice quotes from my blog post explaining what was going on. The problem is that these quotes were not written by me, never existed, and appear to be AI hallucinations themselves.
Nice job, Ars
