I think it goes beyond this. I was just using claude to edit a blog post which mentioned OpenClaw and I got this response: "The "OpenClaw" reference — I assume that's a typo or playful reference; if you mean a real product, I couldn't find it under that spelling and you'll want to fix or footnote it.". I gave it a direct link to openclaw.ai and the chat instantly ended and hit my 5hr usage limit. Could have been a coincidence, but I had only lightly been using sonnet in the morning so it seems unlikely. Very odd.
> I don't know what "openclaw" is. It's not something I have knowledge of, and it doesn't appear in your memory or this project's context.
As others have pointed out, Anthropic is allowed to have TOS, even if we disagree with it.
But having Claude deny the existence of OpenClaw is a way more hazardous and likely straight up violates Claude's Constitution: https://www.anthropic.com/constitution
> As others have pointed out, Anthropic is allowed to have TOS, even if we disagree with it.
Anthropic is allowed to shutdown its LLM and manufacture clown noses if it wants
Doesn’t mean customers have to agree with it.
Come on, folks. This is not a conspiracy.
LLMs have a knowledge cutoff date. Opus 4.7's documented cutoff date is in January. Older Claude models are earlier than that.
OpenClaw didn't have the name OpenClaw until January 30th. So indeed, even the latest Claude model does not know what OpenClaw is, unless you have it do a web search. If you have it search, it'll happily tell you all about it.
Knowledge cutoff is completely insufficient as an explanation.
These models have access to a web search tool. Gemini and ChatGPT both happily search for give info on OpenClaw. Claude denies all knowledge.
What’s more it’s this part that’s very concerning.. Banned for wrong think..
> I gave it a direct link to openclaw.ai and the chat instantly ended and hit my 5hr usage limit.
Claude does not use the web search tool unless it thinks there's a good reason to. If you nudge it to search, it will, and then it'll tell you all about OpenClaw. You can easily go try this yourself -- I just did. It works fine.
> I gave it a direct link to openclaw.ai and the chat instantly ended and hit my 5hr usage limit.
Again, go to claude.ai yourself and try it. It works fine. It happily tells you about OpenClaw.
Whatever happened to jrflo must have been a coincidence. It frankly doesn't make any sense for Anthropic to be trying to block this.
Claude is notorious in my experience to lie directly to your face (even if its baffling) instead of using the web tool. I'll never prompt it without saying 'use web tool' Its ridiculous.
the model has no such magical access to tools.
it's the harness which responds to the models replies that has access to the tools.
I wish people would continue to reiterate this difference.
I said the model has access to a web search tool. Of course it’s the harness that provides that access.
I didn’t spell that out because it’s irrelevant detail, immaterial to the point I was making.
Except GP said they also pointed it to the source website to reference and then had the follow up weirdness.
Is the behavior the same with other unknown words? Certainly doesn't seem so from other comments.
Fair call.
I don't think couching it as conspiracy is the right frame either. This is not a one-off. I think a critical eye is warranted.
> likely straight up violates Claude's Constitution
A company that goes against their self-proclaimed values... What a shocker.
>> likely straight up violates Claude's Constitution > A company that goes against their self-proclaimed values... What a shocker
Makes you wonder how much of the claims around Mythos are exaggerated to crate hype in advance of a IPO
Surely and AI company wouldn't exaggerate. Thats securities fraud!
It doesn't look like anything to me
For those that don’t get this. It’s a reference to West World, where the “hosts” (androids) say this sentence when they see something from the outside world that they are programmed to ignore
Seize all motor functions.
It's not "_cease_ all motor functions"?
I thought it was freeze all motor functions!
I did a quick ctrl+f through the season 1 .srts and it looks like it's usually freeze but sometimes cease. E.g. S01E10 has both in different parts.
The weird thing is that it found sources for all of my other claims and references no problem, but acted like it didn't know what openclaw was when openclaw.ai is the first thing that pops up on google.
"OpenClaw" is a name from January 27, 2026. It's new enough that it's not in the training data for a lot of AI models. So they, quite literally, don't know what it refers to.
"If you don't know an identifier, google it" isn't a very reliable behavior in today's models. They do it, but only sometimes.
That's true, it could have been going from training data and skipping an explicit web search, but it was odd because I specifically asked it to pull references for my blog post, and it pulled ~20 links in the same message it said OpenClaw doesn't exist.
That's not how any of this works.
That's exactly how it works.
Going off-topic now, but you probably would want a "knowledge cutoff date" in Westworld, wouldn't you?
Can't have the Hosts getting riled up about the Gavinite-Baronite skirmishes, even if the Guests are all hot and bothered.
This is some real "There is no claw in ba sing se" stuff.
Dragons steal gold and jewels... and they guard their plunder as long as they live... and never enjoy a brass ring of it. Indeed they hardly know a good bit of work from a bad, though they usually have a good notion of the market value
My theory is the dragons actually benefit immensely from sitting atop the gold piles as it acts as an amazing heat sink.
I don’t think that really fits with the metaphor but I wanted to say my piece regardless.
We don’t really have dwarven gold hoards anymore - I’m thinking we can prove climate change is caused by overheating dragons.
Everyone send me all your gold and I’ll prove it.
Why do you think places like Fort Knox have never been robbed? They have the best security guard.
People I wouldn't focus on heat sinking. I would focus on hoarding! Not letting others to share their precious things with them
I always thought dragons were reptilian and therefore cold blooded.
Yes, but being cold blooded doesn’t mean their blood is actually cold, it just means that they cannot internally regulate their temperature. For the majority of creatures that means they need external sources of warmth, dragons are unique in that they need external sources of “cool”.
Same past days it sometimes tried to gaslight me saying OpenClaw isn't a thing.
This is a death sentence for Anthropic if true.
Trash models that dont represent reality. What else is RLed out
Lmao, I can 100% believe that they are deliberately filling your usage bar to sabotage their competition. These people have no morals.
"Sorry, that was a bug!" Thariq will be on scene shortly, don't worry.
Yeah it will be something like "we A/B tested on 0,05% of users and ..."
I mean that also just sounds illegal...
It also sounds extremely counterproductive to try and sabotage your competition by.. driving your customers away? I have no love for these companies but it's a silly conclusion to jump to.
They don't want customers that make them bleed more money than they're supposed to.
I'm starting to realize that fixed-price agent subscriptions follow the same profit model as gym memberships.
People on OpenClaw discord were bragging about having this stuff running 24/7 and using billions of tokens. I think one guy was using billions per day. (I might have misplaced some zeros but I remember one guy's bill would have been $1000 with API pricing. Per day.)
At the time, enforcement was pretty random, and I think based on how heavy your traffic was.
They weren't all on Claude (though it was the preferred setup) and some people had dozens of accounts hooked up with proxies to avoid hitting limits.
Then just... charge everyone in same way ? The problem is entirely caused by their ass-backwards billing methods
But the have an ass-backwards billing method to appeal to the masses in the first place. It's like price dumping as long as they can do it with the investors' money that they somehow swindled. Their competitors do the same thing, so it is either go along with it, or be left behind in the dust. A contest of endurance in financial swindling.
I for one hope it all comes crashing down, when reality hits these companies. I like being able to ask some LLM a question, when I don't know something. I also like asking it for examples. But I don't let it write my code and burn tokens to no end until it passes some tests or something. My usage is at human speed, and I feel like that is sufficient for the technology to be helpful. For the rest I will use my biological wet ware, thank you.
They're subsidizing the plans. A lot of subscriptions in general do this: the users that barely do anything subsidize the users that do a whole lot. If every user starts doing a whole lot more than usual, you have a problem. Which means OpenClaw poses a problem, because not only do existing users start doing a whole lot more than usual, but a huge influx of new users start doing the whole lot too.
Would they act differently if it was?
Not if a chatbot did it, maybe. No legal precedence here. Also they are a defense and offense contractor they could kill people and nothing would happen
Chatbot doesn't really make a difference. Swap out Claude with the aws or azure cli increasing your usage to 100% for mentioning some forbidden keyword and it's the same problem.
The chatbot didn't do shit? The person coding the chatbot put this explicit condition in place.
I was just using claude to edit a blog post
There's your problem.
I mostly use it to get a general vibecheck, it's pretty decent for fact checking and identifying narrative gaps, as well as finding sources for things I know are true but don't want to spend the time to manually find. Having the LLM output itself get posted is pretty dumb and somewhat disingenuous to people who read it IMO, I'm not just shoveling that out onto the internet
Why not? I do the same, I tell it the exact content, but I don’t have to do all the rest. My blog is a react based (because I like interactivity) and has no asset pipeline, so it’s not as user friendly to edit the content as e.g. a markdown file.
What do you need React for in a blog?
Check this post for example: https://tn1ck.com/blog/how-to-generate-sudokus
I use it for the interactivity.
[dead]