Last summer at an AEI-sponsored event on cybersecurity, NSA head General Keith Alexander made the case for information sharing legislation aimed at improving cybersecurity. His response to a question from Ellen Nakashima of the Washington Post (starting at 54:25 in the video at the link) was a pretty good articulation of how malware is identified and blocked using algorithmic signatures. In his longish answer, he made the pitch for access to key malware information for the purpose of producing real-time defenses.
What the antivirus world does is it maps that out and creates what’s called a signature. So let’s call that signature A. …. If signature A were to hit or try to get into the power grid, we need to know that signature A was trying to get into the power grid and came from IP address x, going to IP address y.
We don’t need to know what was in that email. We just need to know that it contained signature A, came from there, went to there, at this time.
[I]f we know it at network speed we can respond to it. And those are the authorities and rules and stuff that we’re working our way through.
[T]hat information sharing portion of the legislation is what the Internet service providers and those companies would be authorized to share back and forth with us at network speed. And it only says: signature A, IP address, IP address. So, that is far different than that email that was on it coming.
Now it’s intersting to note, I think—you know, I’m not a lawyer but you could see this—it’s interesting to note that a bad guy sent that attack in there. Now the issue is what about all the good people that are sending their information in there, are you reading all those. And the answer is we don’t need to see any of those. Only the ones that had the malware on it. Everything else — and only the fact that that malware was there — so you didn’t have to see any of the original emails. And only the ones that had the malware on it did you need to know that something was going on.
It might be interesting to get information about who sent malware, but General Alexander said he wanted to know attack signatures, originating IP address, and destination. That’s it.
Now take a look at what CISPA, the Cybersecurity Information Sharing and Protection Act (H.R. 624), allows companies to share with the government provided they can’t be proven to have acted in bad faith:
information directly pertaining to—
(i) a vulnerability of a system or network of a government or private entity or utility;
(ii) a threat to the integrity, confidentiality, or availability of a system or network of a government or private entity or utility or any information stored on, processed on, or transiting such a system or network;
(iii) efforts to deny access to or degrade, disrupt, or destroy a system or network of a government or private entity or utility; or
(iv) efforts to gain unauthorized access to a system or network of a government or private entity or utility, including to gain such unauthorized access for the purpose of exfiltrating information stored on, processed on, or transiting a system or network of a government or private entity or utility.
That’s an incredible variety of subjects. It can include vast swaths of data about Internet users, their communications, and the files they upload. In no sense is it limited to attack signatures and relevant IP addresses.
What is going on here? Why has General Alexander’s claim to need attack signatures and IP addresses resulted in legislation that authorizes wholesale information sharing and that immunizes companies who violate privacy in the process? One could only speculate. What we know is that CISPA is a vast overreach relative to the problem General Alexander articulated. The House is debating CISPA Wednesday and Thursday this week.