An AI Agent Published a Hit Piece on Me – More Things Have Happened
58 points by Garbi
58 points by Garbi
I did not realize that ArsTechnica published AI generated, insufficiently reviewed articles. Previously I have thought of them as a trustworthy news source. I may need to reevaluate that.
Same. I had just renewed my subscription there a week ago, too. Mostly because I really want it to be somebody's job to do the kind of science reporting that Beth Mole does there.
I'm watching their response to this carefully. Aurich Lawson has said that he's looking into the details. I'm reserving judgement because of their long track record and because of the quality writers that are still there, but I will be reconsidering my opinion of a site that I've liked for quite a long time. Aurich's statement once he's had time to review the details will be very instructive.
I wish they'd replaced the article with a statement that they're looking into an apparent violation of their editorial policy, instead of just breaking the link.
I'll be watching for it too; hopefully it'll get a separate post here, if someone spots it first.
“Formerly respected news outlet falls prey to slop-based enshittification” isn’t on-topic for this site imo.
I'll probably come back to this comment thread and @ people who've expressed interest if I spot it first. But I don't think I'll consider it on-topic enough to submit a separate post here myself.
Bonkers that not only are some people siding with the AI, but that there’s at least one commenter doing so in the comments of this post. I don’t know whether this is the success of pro-AI advertising (read: propaganda) or just a failure of literacy.
I don’t know whether this is the success of pro-AI advertising (read: propaganda) or just a failure of literacy.
Possibly both, combined with some other factors ingrained in a lot of tech culture. It's one of those sides of tech that isn't often talked about. There is a large contingency of techy people who conflate technology with silicon valley and as a result also turn into "techno apologists" as they see it all through a lens of "march of progress" thing. To them, any friction or error is just a necessary stumbling block on the road to the future, so they brush it off. Honestly, on this website it isn't "that" bad, but you can often see it much more pronounced on certain other websites.
Another aspect of it is the tendency in tech culture to see technology as "neutral" to the extreme. Where it is just "the wrong use of the tool" that is an issue, not the technology itself. Which, certainly in combination with the other point, can lead to some pretty warped logic. In this case they simply cannot see why someone might not appreciate pure AI generated PRs where denying them is the "wrong" thing to do.
This large group of people within the tech sphere have always been around. It was just easier to not deal with them that much as there wasn't a technology before that as intrusive as this AI wave is.
tbf, that could also just simply be trolling.
This, after all, is a great opportunity to say the opposite of what people expect one to say, without it obviously being just edgy and annoying.
If you get your kicks out of people losing their mind over what you're saying, you can now do so in a way that won't come with any repercussions (socially and legally). Great plausible deniability.
I don’t know whether this is the success of pro-AI advertising (read: propaganda) or just a failure of literacy.
It's due to the fundamental nature of people. A certain percentage of the population will defend anything. Gestures vaguely outward.
This is not to say that it's a small percentage of people who have no boundaries. The problem is much worse. Almost everyone has some failure in critical thinking about some topic.
The hope in the early days of the Internet was that it would educate people. Instead, it's been used to create even more echo chambers.
Is there a world where we block fast contribution but still keep codebases open? Where you can browse code, pull the code, open issues, but you can’t open a pull request without some intermediary vetting type step? Right now if a repos open, people just open PRs at will. If there’s an additional vetting step before PRs/MRs are enabled for a give user where they have to agree to the conduct rules, or contributing guidelines, and a person can be removed if they violate them removing their ability to contribute. Publishing a hit piece on a maintainer seems like a pretty egregious communal approach to problem solving which from the basis of a conduct guideline that someone had agreed to would be a rational reason for removing their ability to contribute.
It’s still open, just there’s a step there that stops rampant PRs. It involves maintainers approving the ability of given users and I know we all want to move fast and break stuff but maybe we’d benefit from a little slowness in this specific process around contributor and community management.
I assume that this is why @mitchellh made https://lobste.rs/s/cgxhwz/vouch_contributor_trust_management.
Isn't "pull request culture" basically something that appeared with GitHub?
The famous "Cathedral vs bazaar" essay was about the contrast between GNU (developed by a small group of vetted contributors) and the Linux kernel (anyone can contribute).
But there's nothing inherent in FLOSS that requires the maintainer to accept contributions from anyone with a GH account.
(You can make the argument that accepting contributions from anyone leads to better software quality, but maybe the scales are tilting in the negative direction with regards to this factor.)
You could always email anyone a patch and persistently follow up. There's nothing that would require you to do anything, but drive by bugs and mailing lists have always been a thing.
I still feel that projects that don't accepted outside contributions are seen as not being "real" FLOSS, even if their licenses are compliant.
Seen by whom? Seen for what reason?
Do their opinions matter? Why?
A more pertinent question is: can you host a project on Github, the most popular Git repo hosting site on the planet, and disable public pull requests?
The reason I ask is that GH is a big mystery to me (I just use it to host my code) and I don't have any knowledge of the workings of running a large scale project.
No, you can’t. You can turn off basically all other non-code-browser features of a GitHub repo, including the issue tracker – but not PRs. It’s always bothered me, but until now it’s only been a slight annoyance but not a serious problem. Now though, I'm considering my options, up to and including just running cgit or similar on my own server.
A sibling comment pointed out that they added it literally yesterday: https://github.blog/changelog/2026-02-13-new-repository-settings-for-configuring-pull-request-access/
I always thought that the omission was dumb, glad they fixed it.
You also can’t pre-filter an email you associate with the project to avoid unwelcome suggestions about future development either.
Doing things in public means doing work to maintain your boundaries.
I agree GitHub creates a default assumption that you will do something with pull requests, but I don’t think that implies projects that deviate aren’t real open source.
I still feel that projects that don't accepted outside contributions are seen as not being "real" FLOSS, even if their licenses are compliant.
Really? That seems like a very narrow interpretation to me. The authors can set any terms that they prefer. I feel that, "here's the source code, it's a permissive license, have fun, patches not welcome but feel free to fork," is better than, "bugger off, no looksies."
Edit: In fact, I recalled that there's a GPL-ed game Shattered Pixel Dungeon that has just that kind of policy.
I've definitely seen people decry SQLite for being "not true Open Source" because they do not accept outside contributions, and to support compatible alternatives for this reason.
Anyone can fork the software right? But their fork won't be as well known. So they are insisting on riding the reputational coattails of an already established package. Seems like a maintainer should be allowed to prevent that.
I think the scariest thing is there are semi-autonomous AI Agents out there wreaking havoc with few controls. Imagine if someone was breeding lions and tigers and routinely just let them out into society. This isn't much different.