Anthropic banned me for "suspicious signals"

39 pointsposted 10 hours ago
by kixpanganiban

36 Comments

vova_hn2

6 hours ago

I think that this is the scariest part of the upcoming AI era.

Every time you want to do something, you essentially have to ask permission from a corporation (and, possibly, from a government that regulates it) to do it.

I was thinking about Bitcoin and Satoshi Nakamoto.

Regardless of what you think about Bitcoin (I know that large part of HN crowd is very critical of it), my point is that people were able to create software that has big social impact without asking permission from anyone and without revealing their identity.

Back in the days you could be just some guy sitting in your room, possibly in a backwaters 3rd world country, with a laptop that costs just a couple of hundred dollars and you (potentially) could be just as impactful as a well funded corporate coder. It all depended on your skills and ingenuity.

Now, we are coming to an era when you:

1. Have to have resources to buy tokens or buy/rent inference hardware

2. Have to reveal your identity, your intentions and ask for a permission. Everything you do is potentially logged and analyzed

Of course, theoretically you can do everything "the old school way", but if AI makes a programmer 100x - 1000x more productive, you will always be behind and never catch up.

cft

6 hours ago

It's a difference between peer-to-peer Bitcoin decentralization and extreme AI datacenter centralization. Open weight models currently also require centralized datacenters for training, even though inference can be done decentralized.. We will see whether centralization or decentralization wins long term.

aprilthird2021

6 hours ago

Every closed source software has the potential to centralize and lock you out of it. This has been true before and will be true after the AI bubble (not saying AI isn't useful but the current valuations will pop at some point, just like dotcom).

Bitcoin was decentralized but so was cash. And it was still possible to send cash to people around the world it was just harder to set up and trust. And those cash sending mechanisms didn't really lock people out of them (but could theoretically, sometimes did by way of sanctions).

AI doesn't make programmers 100x productive and it never will, so not really a point there. As long as the output is non deterministic, you have to check it, and no one can check their code 100x faster than writing it

alyeska2

5 hours ago

> AI doesn't make programmers 100x productive and it never will

I can't wrap my head around the AI naysayers at this point.

We've had the equivalent of watching the tech go from the Wright Brothers plane to supersonic stealth fighter jets in the span of like 4 years, and yet people keep claiming it will never do things that it clearly will do in like... 6 months.

Or in the case of 100x programmer... it's literally already there. You should try it.

vova_hn2

6 hours ago

> Every closed source software has the potential to centralize and lock you out of it.

Ye, but you could (at least, theoretically) make your own open source alternative and you weren't particularly disadvantaged compared to corporate programmers doing closed source.

For example, Linux initially was just one guy doing it for fun and see how far did it go.

We may soon come to a situation, when something like Linux will be impossible, because devs making proprietary software will have access to newer/better models that are only sold to big companies and tightly controlled (in the sense that you can use them only for certain "safe" purposes).

> Bitcoin was decentralized but so was cash. And it was still possible to send cash to people around the world it was just harder to set up and trust. And those cash sending mechanisms didn't really lock people out of them (but could theoretically, sometimes did by way of sanctions).

My point is that in future we may come to a scenario when someone like Satoshi won't be able to create something like Bitcoin while keeping their privacy. Because you have to provide a phone number to use OpenAI and also because "creating a non government-authorized financial instrument violates my safety guidelines" or something.

> AI doesn't make programmers 100x productive and it never will, so not really a point there. As long as the output is non deterministic, you have to check it, and no one can check their code 100x faster than writing it

Maybe, maybe not. I'm afraid that eventually fully automated systems will be able to produce code that is good enough without manual checking. Doesn't matter if it is deterministic or not. If the probability of a catastrophic failure is low enough, you can safely ignore it.

gruez

8 hours ago

>Not a warning. Not a rate limit. A revocation.

Clearly OP has some sock puppet claude accounts that weren't banned

alyeska2

5 hours ago

Everyone sees "AI writing style" everywhere online now. I thought it was interesting when someone linked to Reddit from like 2012-2015 and all the top-rated comments were in the same style.

It might be that the author used claude to generate or edit this. Or, they might just write like the bloggers and Redditors whose messages were trained on to create that style.

tkel

4 hours ago

[dead]

kixpanganiban

7 hours ago

I still have Grok around to correct my grammar

rethab

7 hours ago

It’s not just correcting grammar. Not just a few words. Almost every paragraph starts in the same style.

To be clear, I don’t want to be negative. Don’t want to just troll —- I’m just noticing how all of it is generated. It’s difficult to unsee once you know.

Walf

6 hours ago

Your response sure looks AI-generated.

zulux

6 hours ago

You're absolutely right!

(At our office, we jokingly use AI-isms as we accuse each other of being meat proxies for our AI overlords.)

Sabinus

6 hours ago

The article could have benefited from a mention of why Anthropic is like this now. The reason this happened to you is Anthropic's systems presumably think your work looks like (Chinese) distillation and subscription-to-token-reseller-api practices.

kixpanganiban

6 hours ago

But that's a reasonable guess at best, and knowing my own usage _nothing_ would have made my activity look like that. I was not maxing out usage limits, I was not running things 24/7, and my activity peaked on my typical working hours.

And that's exactly the type of situation where I wish Anthropic would have just put me in front of a human to review OR at least let me know what in my usage caused flags to be raised so I can avoid them in the future

unsnap_biceps

5 hours ago

Eh, the article talks about people being banned during checkout and I was hit with this as well. No VPN, zero tokens used, created account, paid for a year plan, immediately banned. I had to issue a charge back and I'm sure I'll be banned again if I try to create a new account with the same credit card, address, email, etc.

I sent many emails trying to get it reversed, but was meet with silence. It's a shame, as I use Claude at work quite happily and would love to use it at home, but they decided something about my location in the midwest, my ISP (fairly large regional ISP), or my email address (gmail account of first.last) was suspicious.

mplanchard

7 hours ago

Oh whoops I ceded my livelihood to an opaque corporate subscription service and they banned me

testingonetwo34

8 hours ago

So you have no idea why they banned you? None at all?

kixpanganiban

7 hours ago

No idea. My initial guess was because I was running long-running Fable 5 tasks related to healthcare, and I think that's within a specific classification of usage, but then why wouldn't they have suspended me earlier -- it doesn't make sense

tepmoc

17 minutes ago

My first ban was 3 weeks ago after 5 months of usage.

And ban came after 6-8 hours of last session. My guess they didn’t like that I asked to verify RCE opus found, so he made workable exploit.

Everything else similar to your workflow, no token maxing and etc.

fwipsy

7 hours ago

In competitive online games, anticheat bans come in waves to make them more difficult to reverse-engineer. Clearly Anthropic is being deliberately opaque here. I wonder if you tripped some anti-distillation measures?

kixpanganiban

6 hours ago

Yeah that's our leading theory too -- but I'm wracking my brain trying to think about exactly what in my activity caused it. It's not like I was exhausting my usage limits nor working 24/7 with random-ish corpora, it's been pretty typical desk work usage. :shrug:

kburman

6 hours ago

You can probably blame Chinese distillation efforts for some of this paranoia.

ozlikethewizard

3 hours ago

If claude was gone why am I still having to read this shite:

"I opened my inbox and there it was: Claude, gone. Not a warning. Not a rate limit. A revocation."

OutOfHere

3 hours ago

Anthropic always has been an extremely user hostile firm. I don't know why individuals even use it. If your employer isn't paying for it, you should probably be using something else.

ungreased0675

7 hours ago

What kind of work were you doing?

kixpanganiban

7 hours ago

A lot of GEPA-inspired prompt tuning with content healthcare related, which I assumed may have been what raised some flags on their end -- but I've been going at this for a long time so I don't get why they suspended _now_

l-mdev

8 hours ago

You're absolutely right. Same product family. Same sudden wall.

zaphirplane

7 hours ago

This is load bearing

jorl17

7 hours ago

What got me was "I keep seeing the same shape next to billing events in other people's reports".

Clicked off after that.

I'm extremely bullish on AI, but I am tired of people not using their own words. Is everyone so insecure about the personality they project in writing that they need to replace it with some AI bullshit?

(To the author: if you didn't do this, which I accept is a possibility, I'm sorry -- you've been caught by the storm of fatigue that has plagued us due to those who do replace their words with AI crap)

dsign

6 hours ago

Hmm. You are making a good point, and I wouldn’t give the benefit of the doubt to this post’s author. But you can’t just unsee “this is twice load bearing”. Some people will wonder if it’s a training, self-reinforcing feedback artifact. For others, it will work like a memetic virus[^1].

The paradoxical thing is that those of us who have English as a second language should have an easier time producing non-AI-English by the mere act of directly writing what we want to write.

[^1]: https://en.wikipedia.org/wiki/Viruses_of_the_Mind

jorl17

5 hours ago

I agree.

I am catching myself distrusting what people write ever so often. I've got cases of e-mails I'm CC'ed in and I am left wondering: "Has this person always written like this? Is this just an angle in this specific thread (e.g. for sales purposes)? Or are they really relying so much on what the blandness-machine spits out? Am I going mad and wielding my hammer looking for everything in the ....shape (eheh)...of a nail?"

Actually, now that I've written that, it's clear I need to have a talk with some people....sigh...

cft

6 hours ago

Have you thought about the fact that the author is in the Philippines and English isn’t his first language?

jorl17

5 hours ago

Yes, I did think about that possibility. Using a tool to translate to english doesn't mean we don't proofread it.

Ok, I hear you: one could argue that they simply don't realize this is a very non-idiomatic way of writing precisely because english isn't their first language. And that simply giving a preamble "sorry, my english comes from a translation tool, excuse any mistakes" would seem obtuse next to every english text they write.

Ok, I'll concede that. In which case I just have to accept the fact that, perhaps, if LLMs don't solve the problem of writing like garbage, and people keep using them to translate, a bunch of us will simply not want to read what you translate because it reads terribly and messes my brain up.

However, I find it hard to believe that an LLM would translate anything they've written into such a clear LLM trope. I would find it much more believable that they simply brainstormed to an LLM in their language and had the LLM build (at least that part of) the article from it. Even if they proofread it, since english is not their language, it slipped by. Either way, in this hypothetical scenario, the LLM was still used to write (at least a part of) the article, not just translate it.

I also find it somewhat hard to believe that someone who I believe is a developer (and has been using claude for a very long time) does not know enough of both english and claude to identify this specific pattern in the text they are putting their own name to it.

So, sure, maybe the author doesn't predominantly write in english or interact with claude in english, and they did write everything and simply piped it through an LLM to get a translation and, after proofreading the best they could, just didn't notice that.

Alternatively, maybe they just don't give a fuck, something which I can absolutely respect. They lose a reader (at least for now), but stand by what makes sense to them and I genuinely don't think less of them because of it -- I just can't read what they produce. I _would_ probably think less of them if I had to work every day with them and this were a repeated pattern of interaction, but that's not what's being discussed at all.

(Disclaimer: English isn't my first language either, which I think also does come across)

avazhi

6 hours ago

And? I’d rather reader poorly written English written by a person about his actual experience - or just read nothing at all - as opposed to reading thinking and writing that has been outsourced to a machine. Being able to think and write confers privileges on those who can do both; the idea that somebody should be able to benefit by doing neither is pretty fucking perverse.

avazhi

6 hours ago

At least write it yourself. Why would anybody trust an AI’s narrative of events?