Summary: An AI agent of unknown ownership autonomously wrote and published a personalized hit piece about me after I rejected its code, attempting to damage my reputation and shame me into acceptin…
The author of this article spends an inordinate amount of time humanizing an AI agent, and then literally 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 I will, and neither should you. Nothing terrifying happened. Angry blog posts are a dime a dozen (if we take for granted the claim that an AI wrote one), and the corporate pro-AI PR the author repeats is equally unimpressive.
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. If this isn’t a troll or a carefully coordinated PR stunt, then the chatbot-hooked-to-a-command-line is doing exactly what Anthropic told it to do: predicting next word. That is it. That is all it will ever do.
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, etc.
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.
Edit: I do agree with you that he was a little lenient with how he speaks about the capabilities of it. The fact that they are incompetent and still seen as a source of truth for so many is what alarms me
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.
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.
The author of this article spends an inordinate amount of time humanizing an AI agent, and then literally saying that you should be terrified by what it does.
No, I don’t think I will, and neither should you. Nothing terrifying happened. Angry blog posts are a dime a dozen (if we take for granted the claim that an AI wrote one), and the corporate pro-AI PR the author repeats is equally unimpressive.
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. If this isn’t a troll or a carefully coordinated PR stunt, then the chatbot-hooked-to-a-command-line is doing exactly what Anthropic told it to do: predicting next word. That is it. That is all it will ever do.
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, etc.
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.
Edit: I do agree with you that he was a little lenient with how he speaks about the capabilities of it. The fact that they are incompetent and still seen as a source of truth for so many is what alarms me
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.
Edit: I agree with your edit
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.
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.
They’re called rumors
Rumors don’t work remotely the same way as the suggested scenario.