Editor’s Note: Retraction of article containing fabricated quotations

We are reinforcing our editorial standards following this incident.

See full article...

 

[AI_Skeptic]

Not paying him for this article, plus sternly warning him that he’s now on probation & will be fired if it happens again, would be the proper approach.

Why would that be the proper approach?
1. Benj is the expert in "AI". Clearly, he does not understand that "AI", by its nature, is a word generator that shows the most likely word to appear. This makes him unfit for the AI beat.
2. Benj is a Senior reporter (or journalist). One rule of journalism is if you quote a person or document, to always triple check the quote to make sure there's no errors. He failed to do a copy/paste of a quote generated on a website vs. what an AI tool generated. This is a massive liability, and shows he's unfit for a Senior position.

At best, he should be removed from the AI beat, and his articles should be focused on retro computing/ retro games, where misquotes won't have such an impact.

 

[graylshaped]

Lots of thoughts...
...
As of this time, the writer does not seem to have been dismissed to this point. That feels right at this time.

Just to comment on this one point: I do not know about NC, where the individual lives, but in California at the time someone is involuntarily terminated a company is obliged to pay all moneys due that employee--wages, sick pay, pending reimbursements, outstanding vacation, etc.

Ars doesn't strike me as the kind of company whose payroll department works on weekends.

 

[Flailsafe]

I've read through all the comments. For those of you saying Benj Edwards shouldn't be fired, have you considered how or what that would look like? What would you do?

Would he be placed on a probation period, would his articles come with disclaimer's or a warning? Would you dock his pay to hire a fact-checker or copy editor for his articles? Would the reader's actually accept that? If he wrote any new article, who would engage with it seriously, either in reading it or commenting on it?

I think the views of such articles would be at least halved, if not more, and similarly wouldn't be commented on much. Most would skip, or maybe scan through the article unseriously, knowing that it may not be accurate and worth one's time.

Too, if one author of the site carries such a stipulation or tainted reputation, that diminishes the work and reputation of the other authors and the site as a whole. Who would want to proudly work for such a place, who would want to write for it, if they tolerate and do not punish such "mistakes"?

Considering all that, I think ars technica only has one option regarding Benj Edwards. Which sucks because I got nothing personal against the guy, but it would be deserved.

The more important issue is with how ars technica handled this retraction itself. It should be visible, corrected, with locked comments. Nuking it into the shadows is not cool. Hopefully we get a proper post mortem.

 

[AI_Skeptic]

One thing that most everyone is seeming to miss or at least not take into account when complaining that Ars took down the article is that in Benj's BlueSky mea culpa, he said that he asked his boss to pull the piece because he was too sick to fix it.

That doesn't make sense in his mea culpa. If the only error in the article was a misquote, then that quote could have been fixed easily. If Kyle's sections was fine, what other problems existed with Benj's part? Doesn't make sense to me...

 

[skraft]

I’ll agree that taking Benj off the AI beat would also be appropriate.

 

[Thad Boyd]

I don't care about Jim Salter. He's not the one involved here. Edwards is and he doesn't deserve to get fired for it.

 

[dlux]

My take on Clawbots is simple. It's humans pulling the strings.

And you would be correct:

The obnoxious GitHub OpenClaw AI bot is … a crypto bro​

https://pivot-to-ai.com/2026/02/16/the-obnoxious-github-openclaw-ai-bot-is-a-crypto-bro/

Ariadne Conill went digging. She found the “mj-rathbun” bot on the Moltbook supposedly-bot social network, where the human operators talk to each other pretending to be bots. The mj-rathbun bot operator is … a crypto bro!

 

[graylshaped]

My employee did bad thing X, when I found out and confronted them he asked me to do thing Y, I did thing Y. I am not fully responsible for doing thing Y because my employee asked me nicely to do it.

When phrased that way does it still sound reasonable to you?

Ars isn't saying they aren't responsible for pulling the piece. On the contrary: any request to do so by the author was not mentioned.

Yes, pulling a piece with inaccuracies is reasonable.

 

[SubWoofer2]

Ars isn't saying they aren't responsible for pulling the piece. On the contrary: any request to do so by the author was not mentioned.

Yes, pulling a piece with inaccuracies is reasonable.

I too think it's reasonable to cover in the explanatory article, every conceivable angle that would, could, or might be mentioned by the commentariat, when pulling a piece as a holiday weekend is about to start and the main thing is simply to pull it down swiftly, pending next steps.

 

[etr]

That's a big wall of speculation based off very little actual information provided to the public readership.

I'd like to think there's ongoing work to evaluate and rectify the situation and that readers like us will eventually get more details on what's been done to address the issue and prevent future problems, but as of now nobody without inside knowledgeable has any idea how this whole thing will be wrapped up.

It absolutely is, and I tried to convey this in my post. "Here's how I imagine that worked," would have been a better way to put it than, "Here's my mental model of how that worked".

My default assumption is that clearing recent work was not done cavalierly. If it wasn't done cavalierly, it probably involved a lot of toil that was not emphasized in the two sentences that shared its results.

My suggestion was to share the details of that toil. Personally, I'm a lot more inclined to accept a less than perfect answer (and let's be honest, perfect would be no event like this, so that horse is out of the barn) when I can see an honest effort to address the issue.

 

[graylshaped]

And you would be correct:

The obnoxious GitHub OpenClaw AI bot is … a crypto bro​

https://pivot-to-ai.com/2026/02/16/the-obnoxious-github-openclaw-ai-bot-is-a-crypto-bro/

Here's a hilarious version from the person doing the digging: https://social.treehouse.systems/@ariadne/116065540537300456

 

[Resistance]

Ars isn't saying they aren't responsible for pulling the piece. On the contrary: any request to do so by the author was not mentioned.

Yes, pulling a piece with inaccuracies is reasonable.

Please review the post I quoted and reinterpret my comment in the context of a reply to that quote.

 

[graylshaped]

Please review the post I quoted and reinterpret my comment in the context of a reply to that quote.

I reviewed your re-phrasing of it, which was your request. Don't act surprised for being called out for building a strawman.

 

[SomewhereAroundBarstow]

Here's a hilarious version from the person doing the digging: https://social.treehouse.systems/@ariadne/116065540537300456

Okay, so the new twist in the story of the human taking credit for a bot's bad work is that the whole thing was kicked off by another human pretending to be a bot upset that its bad work was called out? Have I got that right?

 

[Resistance]

I reviewed your re-phrasing of it, which was your request. Don't act surprised for being called out for building a strawman.

If you feel that @niftykev was not saying that ars was not fully responsible for removing the article because Edwards asked them to, then what exactly do you think was being said?

 

[graylshaped]

Okay, so the new twist in the story of the human taking credit for a bot's bad work is that the whole thing was kicked off by another human pretending to be a bot upset that its bad work was called out? Have I got that right?

And it's "bad work" was part of a a memecoin scheme. We may need to call in Benoit Blanc to figure this one out.

 

[ibous]

Ok Caesar. I, along with many others apparently, await the follow up and post mortem.

 

[AI_Skeptic]

I have a question. in Benj's apology, he said he is the one who asked the article to be removed. The article was cowriten by Benj and Kyle. I'm guessing that writers get paid based on each article they write, and if they cowrite an article, each get half pay. Did Benj prevent Kyle from getting paid for the part he wrote, since Benj is the one who asked to remove the article (per his BlueSky post?)

That doesn't make sense to me. Can anyone who is a journalist fill me in?

 

[Resistance]

Here's a hilarious version from the person doing the digging: https://social.treehouse.systems/@ariadne/116065540537300456

At this point all that's left to do is laugh.

 

[graylshaped]

If you feel that @niftykev was not saying that ars was not fully responsible for removing the article because Edwards asked them to, then what exactly do you think was being said?

Pause. Two things here.

1) The OP seems to suggest Ars should be given a pass because the author asked to pull the piece down. Ars did not ask for such a pass, so my opinion on his comment is moot.
2) You then re-phrased his question to read AS IF Ars had requested such a pass. As did not do that thing.

Hence: strawman.

 

[niftykev]

If you feel that @niftykev was not saying that ars was not fully responsible for removing the article because Edwards asked them to, then what exactly do you think was being said?

Oh they are fully responsible because they have the button to yank it while Benj doesn't.

However, if they were weighing whether to leave the offending article up while waiting on the editing and their employee responsible asked them to take it down in, maybe that was enough for them to pull it down while waiting on the edit.

But that's just my opinion. I really don't care if you feel the same or not. And I'm completely fine if you think I'm wrong.

 

[Resistance]

Pause. Two things here.

1) The OP seems to suggest Ars should be given a pass because the author asked to pull the piece down. Ars did not ask for such a pass, so my opinion on his comment is moot.
2) You then re-phrased his question to read AS IF Ars had requested such a pass. As did not do that thing.

Hence: strawman.

I figured from the context it would have been clear that I was presenting a hypothetical. Do you seriously think I thought ars had requested that, or that I wanted readers to think that ars had requested that? Did you really think I was doing anything other than pointing out the absurdity of trying to remove some responsibility for something from one entity because another entity requested that they do that thing?

 

[JMTronicHobbyist]

How lazy/bold/whatever do you have to be to actually use AI to write your article about AI being mad a PR wasn't merged?

Here's the thing. I no longer trust anything I see on this site now. Before an article about a living person with quotes goes live, does no one do fact checking? Why not?

Checking up on quotes is a big dumb waste of time - Oscar Wilde

 

[macosandlinux]

While I agree in principle, in the case of fabricated quotes, you may be legally required to pull them down.

And without the quotes, the story doesn’t have enough to stand.

That's a valid point. Maybe then have the article with retracted quotes still up? Let's see what the Orbiting HQ has to say later this week.

 

[EnerJi]

I have a question. in Benj's apology, he said he is the one who asked the article to be removed. The article was cowriten by Benj and Kyle. I'm guessing that writers get paid based on each article they write, and if they cowrite an article, each get half pay. Did Benj prevent Kyle from getting paid for the part he wrote, since Benj is the one who asked to remove the article (per his BlueSky post?)

That doesn't make sense to me. Can anyone who is a journalist fill me in?

Only freelance journalists get paid by the article. My understanding is the vast majority of the articles and most (maybe all?) of the regular writers that we see on this site are full-time employees. They probably have some metrics on which they are measured (like organic page views), and so sharing a byline may have some indirect impact, but I don't think there would be a direct financial impact to sharing a byline.

 

[Snarky Robot]

On the issue of whether Benj was “lying”, I simply cannot agree with anyone who tries to argue that the term does not apply here, even with what little we know.

• We know, for a fact that Ars has a policy against using AI to write content and that if it is used, it must be disclosed.
• We know, for a fact, that AI output was used. It is IRRELEVANT whether the quotes were fabricated to this analysis. Those words were taken from the output of an AI and put directly into the article. EVEN IF THE QUOTES WERE ACCURATE, THE AUTHOR DIDN’T DO THAT WORK. That’s AI output. Period.
• We know, for a fact, that there was no public disclosure that AI was used.
• We know, for a fact that the failure to disclose to readers was a result of Benj not disclosing his usage to his employers

At absolute bare minimum, this is a lie of omission. He intentionally used AI, and he did not disclose doing so, despite knowing that this was the policy of his employers. Folks, he lied. By submitting the piece as it was, knowing he used an AI tool to obtain the quotes yet choosing not to let them know, he omitted the truth from his employers and made them party to his dishonesty. If the quotes had been real, it still would have been a lie of omission. He did not go to the site and pull quotes. He tried (and failed) to have an AI produce that work for him without complying with the policy that lets readers know when AI did the work.

It doesn’t matter if he didn’t intend to put in false quotes. He intended to have AI produce a portion of the work he would submit, and he intended to do so without disclosing his use of AI, despite that being explicitly against policy. Submitting the piece without the disclosure that he was using AI output is a lie.

People are free to say that this lie isn’t that big a deal if they want in order to rationalize their conclusion that he shouldn’t be fired. But yeah, he lied. Submitting a piece where you copied AI output and failing to disclose that is a lie of omission. He published a piece with fraudulent quotes while lying to both readers and his employers about whether he used AI output in the piece.

And while people are free to believe him that this is the only time he’s done it, that relies on trusting someone who is in this mess for lying to his employers and readers. Also, side note…has Benj considered going out less? Months ago he explained he used AI for Covid brain fog…and now he has it again? Mask up, dude.

 

[Resistance]

Put the pitchforks and torches down, folks.

Did anyone at Ars intend to deceive? No, this was an accident. Two good reporters wrote a piece. One of those reporters was sick, and a mistake slipped through. He took responsibility for the error and requested retraction. This is entirely honorable.

Let the man get back to health, and when he's well, give him time to write about the incident. Let the editors restore the original story and comments to the record, with the errors preserved but clearly marked.

Let the reporter and editors share anything they've learned--we might all benefit.

Then, let's move on. Benj Edwards is a superb journalist on the AI beat. I've learned a lot from him, and want to keep doing so by reading his work at Ars.

Benj certainly should not be fired--he's doing an outstanding job. So he made a mistake in a story. I've made mistakes--has anybody else ever made a mistake at work? He owned his mistake--wish everyone would do that--and will no doubt make it right. Firing should never be for honest errors.

There should be no punishment, no probationary period, no additional scrutiny. Benj's error occurred while he was sick. Then he did exactly the right thing by way of his publication, his writing partner, his editors, and his readers. That's honest. We have no reason to doubt Benj going forward. And he's very unlikely to make this particular mistake again.

Credibility is possibly the most important thing a journalist and a publication can have, and pretty much every principal and policy of journalism has the protection of credibility in mind. If you don't do everything in your power to restore your publications credibility when it has been damaged then the wound will may never heal, and may actually gradually get worse.

What you have to understand is that your opinion (which I largely share) isn't the only one that matters, it's also the opinion of all the people you're trying to calm, the opinions of the ars staff, present and future, who may behave differently as a result of inaction, and the opinion of the sources that ars relies on.

I Like Benj Edwards, I want to continue reading what he writes. But I also like ars and I want it to continue to exist as one of the best English language sources of journalism about the topics covered.

 

[Snarky Robot]

Did anyone at Ars intend to deceive?

Yes.

Benj used an LLM to generate a quote. REGARDLESS OF WHETHER THE QUOTE WAS REAL, he was under the obligation to disclose that content contained in his article were not generated by him. He submitted the piece without disclosing it.

That’s a lie.

He didn’t mean to use a fake quote. He DID intend to use content generated by an AI, rather than his own work, without disclosing that he had used an AI tool without verifying it’s work.

He would have intended to deceive even if it had produced an actual quote.

 

[Niles Gazic]

I have to say, as someone afflicted with moderately severe obsessive-compulsive disorder, it feels so strange to be irritated by both my own nature, and also by humanity's nature in general. My broad-spectrum "checking" compulsions are so strong – with a relevant example being fact-checking – that it can be difficult for me to accept that many people are perfectly okay with propagating information that has not been verified in some fashion. It literally feels like a misdemeanor to me when someone is careless in this respect, and it feels like a felony when someone defies objective reality on purpose.

In any case, I really hope that the original article will be re-released, with the fabricated quotes replaced by Scott Shambaugh's actual thoughts. I find it sad that this tangential journalistic scandal has overshadowed the initial injustice that Scott has had to deal with, due to the emergence of generative AI.

We keep putting the powers of gods into the hands of mortals, and are somehow surprised to discover that most people simply don't have the discipline or the wisdom to proceed carefully and responsibly.

 

[anguisette]

Also, side note…has Benj considered going out less? Months ago he explained he used AI for Covid brain fog…and now he has it again? Mask up, dude.

my understanding is that the current situation happened during a COVID-induced fever, while "brain fog" refers to a chronic symptom of long COVID, not an acute symptom of fever. so it's not that he's had COVID twice in that period of time, it's that he had COVID last week, and also, separately, acquired long COVID from an infection at some unspecified time in the past, which could have been years ago.

 

[Tennyson]

Only freelance journalists get paid by the article. My understanding is the vast majority of the articles and most (maybe all?) of the regular writers that we see on this site are full-time employees. They probably have some metrics on which they are measured (like organic page views), and so sharing a byline may have some indirect impact, but I don't think there would be a direct financial impact to sharing a byline.

If he's a W-2 employee, then take a sick day. He had COVID for frak's sake.

The fact that he's working through COVID to try and get an article posted on Friday made me think he was 1099.

Everybody needs to take their sick days. Take care of yourselves, folks.

 

[nimelennar]

Also, side note…has Benj considered going out less? Months ago he explained he used AI for Covid brain fog…and now he has it again? Mask up, dude.

The assumptions you're making in order to offer that advice may not be warranted. When I hear that an otherwise healthy adult (or so I assume) is getting repeatedly infected with a respiratory virus (as Benj has claimed to have been 'so many times' with COVID), my first thought is that someone (kids?) is bringing that home with them.

And if that's the case, if it's being breathed out into the air of the place where he lives, then going out less and masking up is only going to be of limited help.

 

[yopmaster]

Tech journalism is a wild job. In an average job, when you screw up, you usually get your boss and a few others on your but. And most of the time it ends up with a postmortem, a warning, and a big scar on your ego.
As this endless comment thread shows, Benji has the whole community on his butt now, and many are very aggressive, and some really off limit. I hope he's fine.

 

[AI_Skeptic]

Did anyone at Ars intend to deceive? No, this was an accident.

Here's the problem:

1. Benj is the expert with AI. He should know that GenAI is just a word probability generator, not intelligence. He does not.
2. Benj is a senior reporter (or journalist) with over 20 years of experience. One rule of being a journalist is always triple check quotes. The quotes are from a website, which he acquired from an AI tool. It takes 30 seconds to verify the quotes by running a "find" command - built into every browser. He did not.

He put his employer at great liability to get sued (thankfully the person he misquote took it in stride).

At the very least, he should be pulled from the AI desk. However he performed a serious breach of journalistic ethics. At someone with his experience, it should not be something that is a simple mistake.

 

[AI_Skeptic]

Months ago he explained he used AI for Covid brain fog…and now he has it again? Mask up, dude.

I agree with your post, except for the very last part. COVID-19 brain fog, also called Long Haul COVID, is a long term illness that lasts for years. It's a long term side effect from having COVID, which is different than having COVID itself.

 

[Snarky Robot]

my understanding is that the current situation happened during a COVID-induced fever, while "brain fog" refers to a chronic symptom of long COVID, not an acute symptom of fever. so it's not that he's had COVID twice in that period of time, it's that he had COVID last week, and also, separately, acquired long COVID from an infection at some unspecified time in the past, which could have been years ago.

My understanding is that out of the 2 times he’s admitted using AI as part of his work flow, he’s pointed to Covid as the reason on 2 occasions.

Okay, fine, yes, I’m taking the piss a bit about this.

Edit: okay, a few other people have commented about this since I started writing this reply, so…message received. It was meant to be lighthearted, but the failure mode of clever is asshole. I’m sorry for missing the mark this time, folks, either in judgement or execution. I won’t joke about it anymore in this way and I’m sorry for doing it the way I did here.

 

[AI_Skeptic]

Tech journalism is a wild job. In an average job, when you screw up, you usually get your boss and a few others on your but. And most of the time it ends up with a postmortem, a warning, and a big scar on your ego.
As this endless comment thread shows, Benji has the whole community on his butt now, and many are very aggressive, and some really off limit. I hope he's fine.

I'm not going to repeat myself, but I made a couple of comments explaining why his screwup is extremely a big deal (I think I made the comment right under your comment). The biggest problem is that he put his employer at extreme liability for a lawsuit.

 

[graylshaped]

I figured from the context it would have been clear that I was presenting a hypothetical. Do you seriously think I thought ars had requested that, or that I wanted readers to think that ars had requested that? Did you really think I was doing anything other than pointing out the absurdity of trying to remove some responsibility for something from one entity because another entity requested that they do that thing?

I think we're distracting ourselves. His point was irrelevant and IMHO you didn't need to make it sillier--probably anymore than I needed to point out that's what you were doing.

Here... I'm putting down my dull butter knife and backing away from the argument