The official response feels AI generated. I suspect this is a preview of our future.

"You're totally right! I'm sorry but you're going to have to piss off anyway. Would you like to spend a few more hours discussing it with our AI chatbot? It won't help. But if it makes you feel better, it will probably cost us an extra $0.12 in tokens."

I'll bet the first human at Anthropic learns about this from HN.

Sounds illegal to me and I'm sure they'd lose in court if you were incorrectly billed for things completely out of your control.

My guess is this response was entirely written by an LLM that is instructed to never to offer refunds or compensation.

Wait, that was the actual response? With the DiCaprio clap? That wasn't a joke?

Refunds and compensation are different though aren’t they? I would not see being refunded for the billing as compensation, compensation would be something more like $x extra to make up for the inconvenience / to say sorry essentially.

Maybe Anthropic is just testing the waters to see what they can get away with. Left unchallenged (court, charge back, whatever) why change course?

> This is very surprising.

Dude what is it with HN and using extra soft words that don't at all mean the actual thing they're supposed to mean.

Nothing there is a surprise.

This is very bullshit and probably (in a better world for sure) very illegal. Can't bill more than you've actually delivered and what the customer in advance agreed on.

Stop with this god-awful corporate-washed lingo. You're not being professional, you're skewing reality.

Not sure that reasoning has ever stood up in court.

The reply looks like it was written by an LLM. Not that this excuses anything.

They’re also objectively not “unable” they are “unwilling” and hiding behind policies as if they are unalterable laws is silly.

> I've never seen a legitimate business not give refunds for technical errors of their own fault.

Granted, it was very much weasel words.

Nonetheless, I read it as they were issuing a refund ("Let me look up your account information to help process your refund request."), but couldn't offer compensation for pain, suffering, loss of use, tracking down the bug, etc.

I could be wrong, of course, precisely because it was (probably AI-generated) weasel words.

This happens even without HERMES.md. This billing cycle my account was billed an extra $200.

I investigated. I was being for a Claude Max gift subscription that has been sent to – what appears to be – a randomly generated 27 char alphanumeric icloud email account that bounces.

Apparently, Anthropic doesn't have a centralized process that allows you to approve, see or revoke "gift cards." And no I can't use this hypothetical gift card. Because I can't see what the system generated, when it generated it, and if the "gift" sent to this 27 character alphanumeric string was redeemed.

Their support bot doesn't work. As it's a possibly suspicious charge (I certainly didn't buy it), I've been trying to get them to revoke it. But the bot passes it to a human and their humans just close the ticket without comment.

I realize that people working at Anthropic are "just" researchers building cutting edge models. And that Claude is really great and all. But hasn't anyone told them about the global legal risk of incorrectly billing millions of people?

What is their legal risk team doing? Their ops team? Or, whoever else is responsible. Even their own models, Opus 4.6, Opus 4.5 and so on will flag this as a legal risk on "max" thinking.

Because even if $200M to $20M seems "insignificant" next to the however many billions they made in the quarter. Knowingly perpetuating fraudulent billing practises is a real legal risk with real prosecutorial (and financial) consequences. It's absurd to me that so much of legal risk analysis fixates on how users use the tools they pay for, but not what's an obvious trigger for class action lawsuits and prosecutorial investigations in most jurisdictions.

This isn't even a threat. The FTC has taken Uber to court, https://www.ftc.gov/legal-library/browse/cases-proceedings/2... and is apparently seeking a few billion in fines?

https://www.independent.co.uk/us/money/uber-lawsuit-fines-bi...

Purposeful unauthorized billing was found to be a scam and defendants were made to fork over assets, https://www.ftc.gov/news-events/news/press-releases/2024/09/...

And this is government action in one jurisdiction. The EU has fines of 6% of global turnover, and yes, they too will seize assets if the fines are unpaid.

What I'm trying to say politely is, does the Anthropic team realize this is an insane legal risk. And to quote Trevor Moore's immortal words, "insanely illegal."

Why would you do this? Does anyone realize the implications of this? At all? Other than the AI models that the humans aren't paying attention to?

screenshots for anyone interested, https://x.com/_areoform/status/2048644232043434354

More and more I feel that the one thing Github needs to turn the tide of bad press, is to allow adding clown or turd reaction emoji on comments

Because it's illegal.

Brought to you by, allegedly, the "Good" AI company.

Well, when your policy is written by an AI, you can get shit like that

In many countries, this also isn't legally tenable.

After going public and getting publicity. You shouldn't have to do that just to get a company to fix their own mistake. They stole $200, where do they get off saying they won't give it back?

Yeah the initial response is stupid but this is getting resolved, not sure where the initial response OP gives in his git issue came from tbh. I only skimmed the git issue, perhaps they clarified.

Because it hit HN frontpage ...

Same.

Back in December the iOS app had a bug ( https://status.claude.com/incidents/6rrnsb1y0kbn) in which buying a subscription thru the Apple App Store would not register with the backend, so you’d be charged but not receive the plan entitlement.

I discovered this because I wanted to upgrade from free plan to the regular plan. I was charged, but remained in the free tier. Thinking it was a temporary bug, I tried buying the max plan. Same result.

I tried cancelling the plan and restarting but I when I went to buy the regular plan again, I was forever tagged as an “Apple” user and so could only manage the billing plan on the iOS app. I tried one more time, same result.

I tried interacting with the support bot and although it agreed that there was a bug and that it should be fixed and I should get a refund, my account never was able to get unstuck nor refunded. I lodged a refund request with Apple, which was relatively quickly refunded. The Bot never did escalate to a human as promised.

Even though the bug was ostensibly fixed, my account (personal email) remains in permanent limbo, unable to upgrade from Free to anything else (I tried again recently and same result - paid but stuck on free plan). I had to create a new gmail just to pay for Anthropic / Claude.

I got given a gift card with around 6 months credit on it. I used up 1 or 2, and last week suddenly the credit disappeared. I reached out through their chat bot, raised a ticket and have been emailing them daily. Nothing. Absolutely not a word. Unfortunately I dont have the option for a charge back.

Once the dispute was resolved on the card side did anthropic claw back the $100? Was your account penalized in anyway?

I once had PayPal refuse to give me my money back (for a delivery) for months even though the postal service status clearly stated: "Address unknown, returning to sender."

I should have denied the PayPal charge on my bank account, that always gets a real human to look into it. Lesson learned.

somehow it's always the expensive path that works fine.

Too much vibe coding

I guess this is what you get when you replace common sense with LLMs.

I get the confusion -- it looks like the reporter of the bug just posted a raw email response that they got without adding any sort of decoration to make it clear it was from an email they got. At least, that's my reading of this.

I'm also not sure if the person/bot who responded was saying "No refund" or that they couldn't issue a refund, or if a Github Issue was an appropriate place to ask for a refund.

Let's hope a human on the other end is reading this and acting accordingly. It all seems like we're only seeing part of a story.

He's the guy who reported the bug. It looks like he copy-pasted an email from Anthropic without context, and the gif is his response.

sasha-id submitted the original bug report, and then bcherny confirmed that it was a bug and that it's been fixed.

Given that, it's almost guaranteed that sasha-id is a legitimate actor.

If you're confused about sasha-id's comment here (https://github.com/anthropics/claude-code/issues/53262#issue...), it's because they just copied and pasted a support response from Anthropic.

I'm confused about the timeline of events; in the PR, the github actions user lists this as a possible duplicate of https://github.com/anthropics/claude-code/issues/53171, which was created earlier, and doesn't seem to be have been edited after the fact. Did sasha-id just copy that bug report and get credit for discovering?

All these claude issues are full of bots, sometimes bots replying to themselves and getting confused. It's impossible to tell what is a real issue and what is hallucination. I'm surprised anthropic even bothers to read them.

In this particular case I think the authors reply is them quoting what support told them?

He is the original author who faced the bug. I believe he just copied the response he received from Antrophic

The second reply post was his copy and paste response from Anthropic's support staff along with a funny meme mocking it. He just didn't put it in a blockquote or quotation marks.

It was obvious to me, but I can see how somebody could get confused from that.

The Keir Starmer of companies

I have a feeling the devs themselves aren't the issue and it probably sucks to have to be the fall guys (though some for sure might buy into all of Anthropic's schemes).

But my best guess is they don't want to put a firm line down because they want to be free to shift it around however they'd like.

> I don't like it, but can't do much about it.

Is the culture really such that you can't escalate an obvious, fairly minor mistake that is turning into disastrous PR?

That would explain a lot of recent Anthropic takes actually.

"opinions are my own"

  - throwaway449933

I’ve stopped using your product entirely. Anthropic may not like it, but I can do something about it.

It reads like the inventors of Claude can't get Claude to apply a "human in the loop" workflow.

You mean you can't do much about it that wouldn't cost your job.

Oh, what I wouldn't give to see the system prompt that tells Claude what it is or isn't "able" to give refunds for. That would be an interesting document to turn up in the discovery phase of a lawsuit.

> I don't like it, but can't do much about it.

"Whether you think you can, or you think you can't—you're right" - Henry Ford

You could quit, for starters

A little human touch goes a long way with customer service and sales. Sorry your management makes you guys look so bad. But yea I am done with anthropic as well. No offense to you all actually making the thing.

Sounds like somebody needs good numbers for IPO

They're making their moves while everyone thinks ChatGPT is shite.

I asked how to get a partial refund (it blew through my quota in a single question) and Claude sent me to Github.

It’s not a billing issue, it’s a bug that leads to the usage of the wrong quota

This case is so easy, a Chinese LLM lawyer would win against it

As I read it, they didn't look up the account to process the refund. They looked up the account to decide whether to process the refund, and then the decision was "no".

The rest of the support response is just pleasantries and padding, to dance around this fact ("Your detailed reproduction steps will be valuable" blah blah).

See also: privacy.com

(Virtual card provider that generates cards as a free-to-the-user service. They make their money from a cut of the standard transaction fees. Cards are locked to a single merchant and it’s easy to configure limits.)

Already way ahead of you. I never started so I consider myself a winner.

On other hand I wonder what other filenames one could include in their repos to cause this sort of behaviour. Kinda a nudge towards people leaving these tools.

I tried to switch to a competing inferencing platform but they have billing issues as well.

I'm in. What's next?

The only revolution that got started in beds successfully so far was the sexual one.