At one point in the article, the author asks Cloudflare's bot if they're launching a Wallet product, and it says no.
> There is no such product in our documentation or dashboard, so treat any email, website, or message claiming to be "Cloudflare Wallet" as a phishing attempt.
What's the point of adding these AI chatbots if they're hopelessly uninformed about your products?
heh yeah I ran into this with one of the few times I used claude desktop. it had no idea what features it had and didn't have, where buttons were in the app, etc. isn't that kind of a core category of knowledge you'd want the chatbot to know?
in my experience, usually it knows this (it is in the system prompt) but it can still get confused. Especially with skills for example, some skills might only work in claude code/outside of sandbox or in desktop but not on web. And it would sometimes not know if it was on the web or desktop.
Most chat support people were contractors hired from third party companies that were given dossiers about their products that were often quite out of date because poor management has always been a thing.
Last time Coinbase had a breach I wrote in to support chat to see what I needed to do. They said there was no breach. I sent them a link to their own blogpost about it. They responded with "wow this is the fist time I'm hearing about this."
> At minimum, a half-way decent support person would ask a few people internally or search Slack before answering.
Of course not. The extremely vast majority of support staff aren't connected to "internal people" and certainly don't have any access to the main company's Slack.
Most of all, those people are paid very little on very tight length-per-interaction targets. They can't spend any time at all looking for stuff outside the docs package or chatting with peeps outside the immediate costaff.
Everything I said applies to overseas labor, too. I’ve worked at both types of companies, those that co-locate support with the product teams, and those that pay bottom dollar and don’t care at all about support.
With the latter type, there’s still almost always a line of communication to corporate. And the support staff still try to help. They are decent human beings, even if the end result kind of sucks.
Its an organic cycle, as the company grows big the number of relevant areas grow too and inter team communication becomes way too costly/impractical. The big company becomes a group of informal small companies, each running in their own direction and at times competing with each other. Areas like customer support are not really good candidate for career growth so they get least resources and manpower.
And as a dark pattern it adds "positive friction" for the company reducing the number of people that will have the motivation of obtaining the real people support.
Several weeks ago I had an issue not being able to login to Verizon's website, so I tried to chat with someone. The chatbot that was gatekeeping was predictably useless said it would redirect me to a human except...it kept prompting me to log in first. It was literally impossible to differentiate from if they literally had no humans online to talk to at all.
I could be the best money saver for them - and ask for a hefty premium for my services - by terminating all support. No costs, nada, full save! Genius, right?! Never gives false info, never!
Ok, ok, need to have a tickmark next to the 'support' item in the quarterlies, let it be an eternal spinning wheel presenting on clicking the 'Our award winning instant support is HERE!' button then. Its close to the real experience anyway, right?
The Cloudflare folks apparently want security issues reported via HackerOne (which wouldn’t let me log in because the Cloudflare CAPTCHA HackerOne uses seems to be broken…).
The takeaway is that everyone makes security hard. Everyone does this anti-pattern of having these other domains that defeat all their own security recommendations.
GitHub for ages had something like githubnext.com where they would make you do this same OAuth dance (except IIRC it was worse - it explicitly said that it WASNT GitHub). Apple has/had an apple.tv microsite or something they hosted content on.
Your bank will send you “legitimate” surveys or communication from some third party domain like qualtropics.com.
This isn't a secfail. Why is pay.cloudflare.com so hard to establish? Why does marketing always get to overpower engineering? I expect Cloudflare services to avoid some sketchy .pay TLD for exactly the reasons this person went through.
An engineer who vibes up a marketing site, and attempts to put it on the same origin as *.cloudflare.com now has to jump through 1,000 hoops of security clearance, customer notifications, etc.
> `pay.cloudflare.com` can't be launched because it doesn't have the proper WAF preventing 25 year old Wordpress exploits, please make sure pay.cloudflare.com/wp-admin.php is blocked. I don't care that it's a Zig application.
I remember just doing SOC2 for a startup and it made just spinning up an EC2 instance require several steps of rigamarole just to be "in-compliance". And if anything goes wrong? Well why didn't you follow the 2,000 step process?
I don't envy anyone who has to deal with issues like these.
Presumably the big scary sysadmins have access to the *.cloudflare.com DNS records, and marketing just needs to push this thing right now and can't wait, so it's easier for them to buy a new domain with a shiny new TLD than wait for pay.cloudflare.com to be authorised.
cloudflare.com/pay probably has a similar chain of approval: if every marketing idea had its own top-level route, it would get pretty crazy with such a big company.
I thought it wouldn't be as I assume :) CloudFlare scans for new tld and either gets in the sunrise period or at a minimum objects to anyone else registering a straight CloudFlare.tld. But CloudFlarepay.com or cl0udflar3.com have more scam risk.
I only realized it wasnt after googling for the phrase “cloudflare.pay” and finding the announcement on Cloudflare’s own blog, which I trust because it is on cloudflare.com
All the bots including Google’s say it’s a phishing scam site probably, since they don’t know Cloudflare has a wallet product.
Ironically, this might be at least partially because the internal security controls at Cloudflare for using or provisioning new domains/subdomains is so difficult and arduous that the team decided the fastest way to go to market is to get an entirely new domain. Possible bonus that the official bug bounty program won't apply either, since it's on a new domain so any vulnerabilities found won't have to be paid out (as much).
Why is this so, so common? They're subdomains. They're free. It's not hitting anybody's budget to publish a new DNS entry. If someone has permission to publish anything in your name, they probably should be able to go make themselves a subdomain.
Because web security has a lot of legacy crap. Such as cookies. Oh you added Domain=foo.com to your cookie? Now marketingvibecodedapp.foo.com can access the cookie from your main domain. What do you mean you thought adding the domain field restricts the cookie to just that domain? Lol no, it's the exact opposite, gotcha, dumbass.
Because they point foo.example.com to AWS. They then let whatever the CNAME is pointing to lapse. Then an attacker registers the lapsed AWS and can now put their content on your trusted domain.
So the solution is to make it even easier to let the domain records lapse by putting it outside the place where everyone would know about it, with the added bonus that now you can super easily let the domain registration itself lapse without noticing so that your customers who have been told to use it can get directly pwned by whoever grabs it?
Which you could mitigate by using a Cert-Policy in DNS and HTST, so an attacker might not get s valid certificate for your domain and at least access is sonewhat restricted.
Counter-argument is: do you really want some team of 90% marketers and PMs throwing up a MVP/WIP codebase for some ancillary product not related to your core business on your core domain? At a minimum you'd want a thorough security review and risk assessment, and that goes against the ethos of "ship fast and pivot as needed".
Why does the TLD matter in this case? How does your security/risk posture change if you launch on myexperiment.mycompany.com vs. www.mycompanyexperiment.website that you had to go out and newly purchase? Asking because I legit don't know.
It looks like they've updated the cloudflare.pay site to link to the blog post on cloudflare.com that introduces wallets. So they fixed it on the same day they launched. That's not too bad, in my book.
I mean, what would stop someone from registering mycloudflare.pay and doing the same thing? Having the link in the other direction seems like what matters more
I guess it's easy to judge from the sidelines but was the screenshot of the site, if not the first tweet, not an obvious scam? And you can say it's from context but I only read the title before my eyes jumped to the screenshot
In the movie Sneakers, a whole scene is taken up sending some guy on a date with Mary McDonnell so she could record clips of his voice. Today she'd just need a phone call or his Instagram. It's getting harder to keep up with who _people_ are online, much less organizations and domain names.
I don't understand what your point is. Do you disagree with any of the concrete suggestions in the blog post about what should have been done differently, or do you think they're hard to follow?
Another entry in "Marketing department starts a promotion campaign for the new product that's indistinguishable from a phishing attack" list. Starting with not using a subdomain on your own, very well-known domain but instead using a completely different one, then not having it shown with the rest of your services on your main web site, et cetera.
Same Story as it ever was. The first time I encountered what I thought was a phishing attack at the bank I worked at 25 years ago, it turned out to be a marketing campaign, with URLs that put our company name as a user before the domain name (back in the day when creds could go in the URL).
It's probably easier for the marketing department to get a new domain up and running that it is for a new subdomain within their own company. Battling Business Units and all that.
You really would think that at least in theory a company like Cloudflare would make it very easy for internal teams to automatically request new subdomains
Running marketing off a separate domain is often a conscious decision because if they start getting blocked for spam, then critical service/operational emails from your actual domain might also get blocked.
Oh good, I'm glad that Cloudflare, proud defender of internet security, is properly focused on the important goal of optimizing for their ability to send promotional emails to my inbox rather than silly things like helping prevent phishing attacks.
Depending on how possible it is for a use case, reducing the attack surfaces and vectors can help, such as being mindful of how much client side javascript exposes anything.
Web security wasn't hard before we started trying to make the web a platform for full executable software.
I never got hacked through the web before JavaScript (never got hacked after either, yet, but it wasn't really possible in the same way to hack someone through the web without some way to execute program logic, which in the old days would have required a much more specific browser exploit to gain RCE).
JavaScript was a mistake. Everything else after that involves "running code in the browser" was a mistake.
Program execution needs to be completely separate from "the web". I don't want any code of any sort running in my browser, at least not any that I don't fully control. "The web" was never designed to be an application platform. It was only designed to be a document platform.
None of this required Javascript. At all. The same potential attack could have been done with good ol' forms. Sure, you think you're signing into "BigBensSuperStore.com", but you're actually handing your credentials right over to "BigBensSuperStore.net".
> "The web" was never designed to be an application platform. It was only designed to be a document platform.
And then it expanded to serve the needs of billions of people instead of the needs of a few researchers. Womp, womp. Get over it, use a JS-free browser to browse your documents, and accept that the world has moved on. Or don't, and rant at clouds, I guess.
Cisco looks to have made money from repression and torture.
Meanwhile a large fraction of neo-nazis, credit card thieves, and DDoS-for-hire sites are on Cloudflare. It takes serious talent (not morals) to attack humanity at scale.
At one point in the article, the author asks Cloudflare's bot if they're launching a Wallet product, and it says no.
> There is no such product in our documentation or dashboard, so treat any email, website, or message claiming to be "Cloudflare Wallet" as a phishing attempt.
What's the point of adding these AI chatbots if they're hopelessly uninformed about your products?
The point is to show how AI you are.
heh yeah I ran into this with one of the few times I used claude desktop. it had no idea what features it had and didn't have, where buttons were in the app, etc. isn't that kind of a core category of knowledge you'd want the chatbot to know?
in my experience, usually it knows this (it is in the system prompt) but it can still get confused. Especially with skills for example, some skills might only work in claude code/outside of sandbox or in desktop but not on web. And it would sometimes not know if it was on the web or desktop.
The next AI benchmark is can an agent understand the product suite of its creators.
The point is to signal to investors that they're all-in on the current fad, thus making the stock price go up.
What’s the point? To save money paying a human to man a support email. That human would have also been hopelessly uninformed for all the same reasons.
> That human would have also been hopelessly uninformed for all the same reasons.
Not really. At minimum, a half-way decent support person would ask a few people internally or search Slack before answering.
In fact, they would have likely already heard about the new product at lunch or something.
Most chat support people were contractors hired from third party companies that were given dossiers about their products that were often quite out of date because poor management has always been a thing.
Last time Coinbase had a breach I wrote in to support chat to see what I needed to do. They said there was no breach. I sent them a link to their own blogpost about it. They responded with "wow this is the fist time I'm hearing about this."
> At minimum, a half-way decent support person would ask a few people internally or search Slack before answering.
Of course not. The extremely vast majority of support staff aren't connected to "internal people" and certainly don't have any access to the main company's Slack.
Most of all, those people are paid very little on very tight length-per-interaction targets. They can't spend any time at all looking for stuff outside the docs package or chatting with peeps outside the immediate costaff.
Everything I said applies to overseas labor, too. I’ve worked at both types of companies, those that co-locate support with the product teams, and those that pay bottom dollar and don’t care at all about support.
With the latter type, there’s still almost always a line of communication to corporate. And the support staff still try to help. They are decent human beings, even if the end result kind of sucks.
Its an organic cycle, as the company grows big the number of relevant areas grow too and inter team communication becomes way too costly/impractical. The big company becomes a group of informal small companies, each running in their own direction and at times competing with each other. Areas like customer support are not really good candidate for career growth so they get least resources and manpower.
If you have no training or knowledge-base to search, sure. But then you'd be an awful support-team employer.
I mean, yes? That sounds pretty accurate for most companies before chatbots became the new hot thing
Uninformed chatbot replaces uninformed person. And the chatbot doesn't need bathroom breaks or health insurance.
And as a dark pattern it adds "positive friction" for the company reducing the number of people that will have the motivation of obtaining the real people support.
Several weeks ago I had an issue not being able to login to Verizon's website, so I tried to chat with someone. The chatbot that was gatekeeping was predictably useless said it would redirect me to a human except...it kept prompting me to log in first. It was literally impossible to differentiate from if they literally had no humans online to talk to at all.
I could be the best money saver for them - and ask for a hefty premium for my services - by terminating all support. No costs, nada, full save! Genius, right?! Never gives false info, never!
Ok, ok, need to have a tickmark next to the 'support' item in the quarterlies, let it be an eternal spinning wheel presenting on clicking the 'Our award winning instant support is HERE!' button then. Its close to the real experience anyway, right?
To avoid wasting customer time, simply make that button close the window.
Web Developers, please follow every best practice, I’m begging you
Marketing people just make bunch of marketing domains. Business people push all kind of BS ideas.
No one is asking Web Developers about their opinion man.
STOP making everything developers fault.
Who made the website.
Who do we call? CTOs I guess.
My main takeaway from this is not that "security is hard" but that cloudflare is pretty incompetent.
The takeaway is that everyone makes security hard. Everyone does this anti-pattern of having these other domains that defeat all their own security recommendations.
GitHub for ages had something like githubnext.com where they would make you do this same OAuth dance (except IIRC it was worse - it explicitly said that it WASNT GitHub). Apple has/had an apple.tv microsite or something they hosted content on.
Your bank will send you “legitimate” surveys or communication from some third party domain like qualtropics.com.
Fricken Proton has a separate domain that lists all of their apps, https://protonapps.com/. This absolutely screams "scam", but no, it's real.
Ffs, just put this on apps.proton.me or something so I actually know it's real!
So it's not just FedEx[1] who does that, but also one of the most important tech company…
[1]: https://www.troyhunt.com/thanks-fedex-this-is-why-we-keep-ge...
This isn't a secfail. Why is pay.cloudflare.com so hard to establish? Why does marketing always get to overpower engineering? I expect Cloudflare services to avoid some sketchy .pay TLD for exactly the reasons this person went through.
>Why is pay.cloudflare.com so hard to establish?
An engineer who vibes up a marketing site, and attempts to put it on the same origin as *.cloudflare.com now has to jump through 1,000 hoops of security clearance, customer notifications, etc.
> `pay.cloudflare.com` can't be launched because it doesn't have the proper WAF preventing 25 year old Wordpress exploits, please make sure pay.cloudflare.com/wp-admin.php is blocked. I don't care that it's a Zig application.
I remember just doing SOC2 for a startup and it made just spinning up an EC2 instance require several steps of rigamarole just to be "in-compliance". And if anything goes wrong? Well why didn't you follow the 2,000 step process?
I don't envy anyone who has to deal with issues like these.
Presumably the big scary sysadmins have access to the *.cloudflare.com DNS records, and marketing just needs to push this thing right now and can't wait, so it's easier for them to buy a new domain with a shiny new TLD than wait for pay.cloudflare.com to be authorised.
cloudflare.com/pay probably has a similar chain of approval: if every marketing idea had its own top-level route, it would get pretty crazy with such a big company.
Google manages it though. And you could have a labs.cloudflare.com/idea and make it easy to add new ideas internally.
What a ride of a read. I was 100% it was phishing and I got really surprised to find out it wasn't.
I thought it wouldn't be as I assume :) CloudFlare scans for new tld and either gets in the sunrise period or at a minimum objects to anyone else registering a straight CloudFlare.tld. But CloudFlarepay.com or cl0udflar3.com have more scam risk.
I only realized it wasnt after googling for the phrase “cloudflare.pay” and finding the announcement on Cloudflare’s own blog, which I trust because it is on cloudflare.com
All the bots including Google’s say it’s a phishing scam site probably, since they don’t know Cloudflare has a wallet product.
Ironically, this might be at least partially because the internal security controls at Cloudflare for using or provisioning new domains/subdomains is so difficult and arduous that the team decided the fastest way to go to market is to get an entirely new domain. Possible bonus that the official bug bounty program won't apply either, since it's on a new domain so any vulnerabilities found won't have to be paid out (as much).
Why is this so, so common? They're subdomains. They're free. It's not hitting anybody's budget to publish a new DNS entry. If someone has permission to publish anything in your name, they probably should be able to go make themselves a subdomain.
Because web security has a lot of legacy crap. Such as cookies. Oh you added Domain=foo.com to your cookie? Now marketingvibecodedapp.foo.com can access the cookie from your main domain. What do you mean you thought adding the domain field restricts the cookie to just that domain? Lol no, it's the exact opposite, gotcha, dumbass.
Because they point foo.example.com to AWS. They then let whatever the CNAME is pointing to lapse. Then an attacker registers the lapsed AWS and can now put their content on your trusted domain.
https://aws.amazon.com/blogs/security/threat-tactic-spotligh...
(AWS have since fixed this problem, but it exists on other services.)
So the solution is to make it even easier to let the domain records lapse by putting it outside the place where everyone would know about it, with the added bonus that now you can super easily let the domain registration itself lapse without noticing so that your customers who have been told to use it can get directly pwned by whoever grabs it?
Which you could mitigate by using a Cert-Policy in DNS and HTST, so an attacker might not get s valid certificate for your domain and at least access is sonewhat restricted.
Counter-argument is: do you really want some team of 90% marketers and PMs throwing up a MVP/WIP codebase for some ancillary product not related to your core business on your core domain? At a minimum you'd want a thorough security review and risk assessment, and that goes against the ethos of "ship fast and pivot as needed".
Why does the TLD matter in this case? How does your security/risk posture change if you launch on myexperiment.mycompany.com vs. www.mycompanyexperiment.website that you had to go out and newly purchase? Asking because I legit don't know.
One example: a subdomain like experiment.example.com can access cookies for example.com.
I mean, it sounds like they're already doing that, just with extra downside
I’d argue this is exactly why this happened and why the poster is correct but missing the point:
Your organizational management is the problem not the technology
If you can’t coordinate internally to roll out a proper domain then I question how well your teams are managed
It looks like they've updated the cloudflare.pay site to link to the blog post on cloudflare.com that introduces wallets. So they fixed it on the same day they launched. That's not too bad, in my book.
I mean, what would stop someone from registering mycloudflare.pay and doing the same thing? Having the link in the other direction seems like what matters more
By "the same thing", you mean writing the blog post on Cloudflare's official blog?
I guess it's easy to judge from the sidelines but was the screenshot of the site, if not the first tweet, not an obvious scam? And you can say it's from context but I only read the title before my eyes jumped to the screenshot
I just read the rest of the article and I'm back with my tail between my legs. I guess I made the author's point.
Don't worry, I think everyone probably went on the same roller coaster with this one
In the movie Sneakers, a whole scene is taken up sending some guy on a date with Mary McDonnell so she could record clips of his voice. Today she'd just need a phone call or his Instagram. It's getting harder to keep up with who _people_ are online, much less organizations and domain names.
Identity is hard y'all.
I don't understand what your point is. Do you disagree with any of the concrete suggestions in the blog post about what should have been done differently, or do you think they're hard to follow?
I tried to signup and got an "Internal Server Error" post the auth callback. Embarrassing for Cloudflare. Par for the vibe coded culture I guess.
Another entry in "Marketing department starts a promotion campaign for the new product that's indistinguishable from a phishing attack" list. Starting with not using a subdomain on your own, very well-known domain but instead using a completely different one, then not having it shown with the rest of your services on your main web site, et cetera.
Same Story as it ever was. The first time I encountered what I thought was a phishing attack at the bank I worked at 25 years ago, it turned out to be a marketing campaign, with URLs that put our company name as a user before the domain name (back in the day when creds could go in the URL).
Fun fact: still can in Chromium-based browsers. https://textslashplain.com/2023/03/22/attack-techniques-spoo...
It's probably easier for the marketing department to get a new domain up and running that it is for a new subdomain within their own company. Battling Business Units and all that.
That's a problem that should be solved then, because literally everyone loses when it's done this way
You really would think that at least in theory a company like Cloudflare would make it very easy for internal teams to automatically request new subdomains
Running marketing off a separate domain is often a conscious decision because if they start getting blocked for spam, then critical service/operational emails from your actual domain might also get blocked.
Oh good, I'm glad that Cloudflare, proud defender of internet security, is properly focused on the important goal of optimizing for their ability to send promotional emails to my inbox rather than silly things like helping prevent phishing attacks.
this is the correct take
Depending on how possible it is for a use case, reducing the attack surfaces and vectors can help, such as being mindful of how much client side javascript exposes anything.
Cloudflare seems to be trying to do EVERYTHING.
I saw the whole Cloudflare Pay thing and had the exact same thoughts - this has to be some sort of phishing...
1. Why is this website blocked when I try browsing it using Brave?
2. Why on earth would you want a financial product from a WAF?content delivery company?
You didn't read the whole article; it's not a scam, it's a new official cloud-flare product.
Web security wasn't hard before we started trying to make the web a platform for full executable software.
I never got hacked through the web before JavaScript (never got hacked after either, yet, but it wasn't really possible in the same way to hack someone through the web without some way to execute program logic, which in the old days would have required a much more specific browser exploit to gain RCE).
JavaScript was a mistake. Everything else after that involves "running code in the browser" was a mistake.
Program execution needs to be completely separate from "the web". I don't want any code of any sort running in my browser, at least not any that I don't fully control. "The web" was never designed to be an application platform. It was only designed to be a document platform.
None of this required Javascript. At all. The same potential attack could have been done with good ol' forms. Sure, you think you're signing into "BigBensSuperStore.com", but you're actually handing your credentials right over to "BigBensSuperStore.net".
JavaScript (and other forms of executing logic within the browser) have made the situation worse, though.
To me, there's a big difference between a domain misread and actively malicious code running in the browser context as a design point.
If a malicious site gets your password, I'm not sure why it matters whether it happened in the frontend or not.
> "The web" was never designed to be an application platform. It was only designed to be a document platform.
And then it expanded to serve the needs of billions of people instead of the needs of a few researchers. Womp, womp. Get over it, use a JS-free browser to browse your documents, and accept that the world has moved on. Or don't, and rant at clouds, I guess.
this from a company whose main product is (was) security.
I feel there's a generalized decrease in quality in software in general.
Cloudflare is your favorite company and they are geniuses?
Dear Diary,
Today my fanboy bubble was burst.
Signed,
Author
Note that I said: "One of my", and Cloudflare has hired a HUGE percentage of the best networking talent I've encountered.
Another company named Cisco used to do that. They built the Great Firewall of China. Hiring talent does not equal good company.
I hadn't read that so I looked it up to verify, and it appears true:
https://www.eff.org/deeplinks/2016/04/ciscos-latest-attempt-...
Cisco looks to have made money from repression and torture.
Meanwhile a large fraction of neo-nazis, credit card thieves, and DDoS-for-hire sites are on Cloudflare. It takes serious talent (not morals) to attack humanity at scale.
Just use LLMs. They can apparently doing everything and all the things