Phishing Detected! warning in Chrome

Question

I have encountered the "Phishing Detected" warning in Chrome browser on my dev site. Interestingly I don't encounter the same warning in Firefox or Safari even though, as far I can tell, they are using the same phishing database (although in Safari preferences it says "google safe browsing service is unavailable"). I also don't encounter the warning on the same page of the production sites.

It first popped up on a new account verification page I created which amongst other things asked users to confirm their PayPal account with the GetVerifiedStatus API. This requires only name and email.

I have also encountered the warning on a configuration page which asks for the PayPal email address which the user wishes to receive payments to.

Neither page requests a password or any other data that would be considered a secret.

As you might gather I have zeroed in on a potential false positive on the PayPal portion of the content as if perhaps I am phishing for PayPal information beyond the payers email address. There has been no malicious code injection or any such thing. Even when i've removed all content from the page the warning is still present.

I reported the first incorrect detection to Google, and intend to do the same for the second incident, however what I really want to clear up is:

1. What content can lead to this warning?
2. How can I avoid it in the future?
3. How can I get some info from the "authorities" on which urls are blocked? (Webmaster Tools is not showing warnings for the dev site)
4. How can I flush my local cache of "bad sites" in case I want to re-test?

Clearly having a massive red alert presented to a user on a production site would be disastrous, and there is a (perhaps deliberate) lack of information about how this safe browsing service actually works.

Answer 1

I have been developing a website for a banking software developer and ran into the Phishing warning as well. Unlike you I had no PayPal associations in any of my code and well not even any data collection besides a simple contact form. Here are some things I managed to figure out to resolve my false positive warnings.

1) The warnings in Chrome (red gradient background) is a detection method built into the Chrome browser itself and it does not require to check any blacklists to give the warning. In fact Google themselves claim that this is one of the methods that they discover new potentially harmful sites. When your site is actually on the blacklists you get another red warning screen with diagonal lines in the background. This explains why you only see the warning in Chrome.

2) What actually triggers this warning is obviously kept kind of hidden. I could not find anything to help me debug the content of my site. You have pretty much done this, so for anybody else in need of help, I had to isolate the parts of my site to see what was triggering the warnings. Due to the nature of the site I was working on it turned out to be the combination of words and phrases in the content itself. (e.g Banking Solutions, Online Banking, Mobile Banking). Alone they did not trigger anything but when loaded together chrome would do its thing. So I'm not sure what your triggers are or even what the list of possible triggers are. Sorry...

3) I found that simply quitting Chrome completely and restarting it resets the "cache" for whether it has perviously detected a page. I closed Chrome hundreds of times while getting to the bottom of my warnings.

Thats all I have and hope it helps.

Update: My staging area was accessed via an IP address. Once I moved the site to use a domain instead all the warnings stopped in chrome.

Answer 2

I experienced the same today while creating an SSL test report for my web server customers. What I had there was simply something like this:

"Compare the SSL results of our server to the results of a well-known bank and its Internet banking service". I just wanted to show that the banking site had grading B whereas ours had grading A-.

I had two images from SSL-Labs (one the results for my server and the other the results of the bank). No input fields, no links to any other site and definitely no wording about then name of the bank.

One h1, two h2 titles and two paragraphs plus two images.

I moved the HTML to the page and opened it in my Chrome browser. The web server log told me that a Google service had loaded the page after 20 seconds from my first preview. Nobody else had seen it so far. The phishing site warning came to me (webmaster) in less than an hour.

So it seems to me that the damn browser is making the decision and reporting to Google which then automatically checks and blocks the site. So the site is being reported to Google by Google tools, the trial is run by Google and the sentence is given by Google. Very, very nice indeed.