Insider Info on the FBI's Spyware

Tucked into an affidavit filed by an FBI agent last month was the first hard evidence that federal agents are equipped with more than automatic pistols and handcuffs: The agency was asking a federal judge to let it infect a PC with spyware so they could finger its owner.

The case, which was reported locally in Olympia, Wash., last month and received more national exposure this month, involved bomb threats e-mailed to Timberline High School in Lacey, Wash., an IP trail that went cold in Italy and a call to the FBI.

Special Agent Norm Sanders, who swore out the affidavit, could be Efrem Zimbalist Jr.'s doppelganger for all we know, but he must have been more talkative than the close-lipped character from the late-1960s TV drama The FBI to win over a judge. Sanders had to spill some beans about CIPAV, the agency's name for what the rest of us would call spyware -- software the FBI wanted to plant on the PC used to e-mail the bomb threats in the hope of identifying its owner, and thus the sender.

Until Computerworld's Freedom of Information Act (FOIA) request is granted and more information on CIPAV is reviewed -- and maybe not even then -- all we have to go on is this:

What is CIPAV? CIPAV, which stands for "Computer and Internet Protocol Address Verifier," is secret surveillance software that the FBI used last month to help identify whoever was e-mailing bomb threats almost daily to a Washington high school. Although at least one security professional agreed that CIPAV fits the description of spyware, much of what it is, or does, is unknown. What is known: The software collects a wide range of information from the target PC and sends it back to control -- in this case, the FBI -- and automatically records every outbound communication, though not the contents of said communication. If that sounds like a bot, well. ...

What does CIPAV do? As the affidavit spelled out, "the exact nature of [CIPAV] commands, processes, capabilities and their configuration is classified as a law enforcement-sensitive investigative technique," so not all the facts are in.

But according to the court filing, this is what the CIPAV collects from the infected computer:

-- IP address

-- Media Access Control address for the network card

-- List of open TCP and UDP ports

-- List of running programs

-- Operating system's type, version and serial number (in Windows, the serial number is the 25-digit alphanumeric product activation key)

-- Default browser and its version

-- Default language of the operating system

-- Currently logged-in user (username) and registered company name (The latter is optional in Windows.)

-- Last visited URL

Once that initial inventory is conducted, the CIPAV slips into the background and silently monitors all outbound communication, logging every IP address to which the computer connects, and time and date stamping each. The affidavit called this a "pen register." The content of each communication -- the data packets that made up an e-mail message, for instance -- were expressly not to be collected.

What happens to the data the CIPAV collects? According to the warrant application, the CIPAV transmits the information to a computer "controlled by the FBI" in the jurisdiction of the U.S. District Court's Eastern District of Virginia. Presumably, the server is at the FBI Academy in Quantico, Va., which is within the Eastern District.

Does the CIPAV capture keystrokes? We don't know, and the FBI isn't talking.

Can the CIPAV spread on its own to other computers, either purposefully or by accident? Does it erase itself after its job is done? We don't know. The only clue in the affidavit is that the CIPAV would operate as a pen register for up to 60 days after the software had been "activated" by the recipient. In other words, the FBI swore that the monitor would "time out" after 60 days. But not that it would delete itself or not be able to spread in some worm or bot fashion.

Speculation ahead: The affidavit was mum on whether the CIPAV collected the kind of information necessary to propagate, such as e-mail addresses in the PC's address book, instant messaging contacts or even, since it was launched at an as-then-unidentified MySpace user, MySpace's messaging list).

Does the FBI have just one stock CIPAV model? The affidavit does seem to hint that the spyware comes in more flavors than just vanilla. It said, "Because the FBI cannot predict whether any particular formation of a CIPAV [emphasis ours] to be used will cause a person(s) controlling the activating computer to activate a CIPAV, I request that this Court authorize the FBI to continue using additional CIPAV's in conjunction with the target MySpace account (for up to 10 days after this warrant is authorized), until a CIPAV has been activated by the activating computer."

How did the CIPAV get onto the targeted computer? Hard to say specifically, but we can deduce some things from the affidavit and MySpace, which the CIPAV took aim at. Some user action was clearly required to infect the PC with the CIPAV. In the warrant application, the FBI used the term activate several times and alluded to a spyware plant failure if the target did not trigger the CIPAV through the targeted MySpace account.

MySpace accounts can't receive traditional e-mail, so one hacker standard -- attach the CIPAV to a message and hope the recipient is stupid enough to launch it -- wasn't available. Instead, the most likely tactic would have been to send a URL to the suspect account using MySpace's own instant messaging and/or Web mail system. If the suspect clicked on the link -- it would have had to be enticing, so use your imagination here -- and visited the FBI-owned malicious site, an exploit for a zero-day vulnerability (or unpatched one on the suspect's PC) would have let the government download CIPAV to the target hard drive.

But which vulnerability? We don't know. Conceivably, it could have been the FBI's own super-duper flaw, but Occam's razor says it was probably one of the many effective, yet run-of-the-mill, bugs in the wild. Roger Thompson, chief technology officer at Exploit Prevention Labs, took a guess. "If I had to bet, I'd bet on ANI," Thompson said in an IM interview.

Good bet. The animated cursor flaw harks back only to late March, and although Microsoft Corp. patched it in an out-of-cycle update on April 2, it's effective enough to still be used by the notorious multistrike hacker exploit kit Mpack as recently as last month, long after CIPAV was deployed.

I remember something about the FBI having something called "Magic Lantern." Any connection? Unlikely, other than as descendant. Magic Lantern was the code name given to FBI-made surveillance software in a November 2001 story broken by MSNBC.com, which outlined a keylogger-type Trojan horse to be delivered as an e-mail attachment.

But that was nearly seven years ago. To give you an idea, that news preceded major security events such as the Slammer and MyDoom worms (2003 and 2004, respectively) and the rise of phishing attacks. Government bureaucracy may move slowly, but seven-year-old security or exploit technology is nearly worthless.

Did the CIPAV work? Apparently. Before the CIPAV's appearance, bomb threats had been received by the school and school administrators on June 4, 5, 6 and 7. Until at least June 8, local police and the FBI had been stymied in their attempts to identify the sender using more traditional methods, such as requesting user information from Google Inc. and MySpace.com and contacting Italian police with a request to locate the computer routing through an Internet service provider's server there.

Once the CIPAV made an entrance, however, the case moved quickly. The warrant application was filed June 12, a Tuesday. At 2 a.m. Thursday, June 14, Lacey, Wash., police arrested an unnamed teenager in his home. The suspect, who had already been identified in news reports as a Timberline High School student, had bail set at US$100,000 in a hearing the following Monday, June 18. On July 15, after he pleaded guilty in juvenile court to charges of identity theft and making bomb threats, the teen was sentenced to 90 days' detention.

With the exception of the affidavit filed by Sanders, however, authorities remained mum throughout as to the specific part that the CIPAV played. We don't know, for instance, when the spyware was activated, whether it was activated after just one version of the CIPAV had been delivered, or what information it collected actually led the police to the boy's home.

Subscribe to the Security Watch Newsletter

Comments