New RFID Attack Opens the Door

Be careful of who walks up to your building and swipes an ID card: New proof-of-concept code will soon be released that lets attackers hack RFID readers and walk right in as if they work there.

The attack uses SQL injection to fake the back-end RFID reader into admitting the cardholder into the building, says Joshua Perrymon, hacking director for PacketFocus Security Solutions and the researcher who wrote the POC. Perrymon -- who's taking a cue from the recent Black Hat RFID flap and won't name RFID vendor names -- says he's tested it on a few RFID vendors' systems, but the exploit will work on most any of them. (See HID Lists RFID Security Steps, HID, IOActive Butt Heads Again, and Black Hat Cancels RFID Demo.)

The RFID databases don't validate the input they receive from the swiped cards, he says, which leaves them wide open for hacks. "I was noticing the back-end database is the same across all products -- I haven't seen any using input validation" to confirm the data they've swiped is legitimate, he says. "It doesn't really matter who the vendor is... In any building you go to with this, bang, you gain access."

Perrymon was able to fake out the readers by injecting SQL characters that appear to be legit into various brands of 1356 Mhz RFID cards. The SQL injection code looks legit when an intruder swipes his card, so he gains entry into the building. "In the user-data section, it uses numeric characters, but we're using brackets and SQL statements... That's standard with a SQL injection in an application."

Conventional attack methods on RFID such as SQL injection haven't been studied much so far, Perrymon says. Most of the attention has been on cracking RFID cryptography and RFID cloning, such as IOActive's research, which was yanked from the Black Hat DC briefing agenda after threats of a patent lawsuit by RFID vendor HID.

Perrymon used an RFID writer to copy a SQL injection statement to the card. Unlike cloning, which copies the user's ID and facility code, this attack uses SQL injection code. "The beauty of this is I'm using off-the-shelf stuff, and there's no reverse-engineering."

Perrymon says adding input validation to these products would be simple for RFID vendors, and he's hoping his work will pressure the vendors to fix the problem. "I want vendors to put in input validation in the reader or database," he says. "Preferably the reader."

But this attack is not for any script kiddie. "You have to be pretty skilled in RFID to understand all the components," says Perrymon, whose company does penetration testing and social engineering exploits. He plans to release the POC soon.

— Kelly Jackson Higgins, Senior Editor, Dark Reading