By Jessica Klein
A video of Nancy Pelosi swept the internet the week before Memorial Day. It depicted the House Speaker slurring her words during a speech she gave at the Center for American Progress Ideas Conference on May 22, in which she criticized President Donald Trump for “obstructing justice.” By May 23, a version of the video posted to Facebook by the conservative Politics Watchdog page had been shared more than 45,000 times and viewed more than 2 million times.
The video had been slowed to three-fourths the speed of Pelosi’s actual speech, making her appear, frankly, drunk. Still, masses took the altered clip as real.
This is the sort of challenge video security companies and law enforcement face in the deepfakes era. When video is so easy to manipulate and disseminate, security providers must come up with new ways to ensure the veracity of all kinds of footage, from CCTV to police body cameras. One solution is to supplement video footage with some kind of indelible mark or immutable code that, when altered with, will definitively show a video has been compromised.
Blockchain technology can do this. Take the bitcoin blockchain—every bitcoin transaction that happens on the blockchain gets a unique hash value, or a “digital fingerprint,” as described by David Jevans, CEO of blockchain forensics and security firm Ciphertrace. This makes the path of a bitcoin entirely traceable. What if the same technology was applied to video?
“You could create a digital signature on [a video segment] the same as you do with bitcoin,” says Jevans, whose company serves clients in government agencies, law enforcement, hedge funds, and banks.
Ciphertrace doesn’t do video forensics, but other blockchain companies do. In 2016, Factom started testing blockchain technology for data security with the United States government. In June 2018, the Department of Homeland Security awarded Factom $192,380 to use its technology at the U.S./Mexico border. Cameras integrated with Factom technology automatically store unalterable information about video footage on a blockchain. If someone attempts to hack into those cameras, the original footage will have already been marked as the real version and safely stored.
Case studies from these efforts haven’t been made public, but other law enforcement offices have successfully used elements of blockchain to preserve video chain of evidence. Take the police force in Northumbria, U.K.
“Previously, the video support unit [in Northumbria] was overloaded with 7,000 video review jobs per year,” says Mark Sugrue, chief technology officer at Kinesese, a Dublin-based video analytics company that specializes in forensic video investigation, with most of its clients in law enforcement. Northumbria police then started using Kinesense’s platform to view and search CCTV footage, which also logs exactly who does what and when with a video using blockchain-like hashing.
“When the user imports any video evidence into the system, a hash of every frame is computed and stored,” says Sugrue. This automates chain of evidence collection, streamlining the process and ultimately cutting down the number of staff needed to review video. Now, the Northumbria Digital Forensic Team reports just 600 video review jobs per year.
Kinesense’s video product doesn’t store these hashes on a blockchain—yet. “I like to say we are ready to provide a full blockchain system as soon as our users are ready for it,” says Sugrue.
There are other inventive ways to use blockchain to track video footage. Custos, a “content protection” company targeting the media industry, has been hiding bitcoin keys in advance film copies for years. The hidden bitcoin incentivizes pirates to seek out those keys. When they find them, Custos knows, and can trace the bounty claim back to the original leak.
Then there’s TruePic, a consumer-facing company that stores image or video data (time, date, location, pixelation) in a hash on the bitcoin blockchain. Ideally, this information is captured the moment the video is taken and stored immediately as part of a tamper-proof record.
Though these last two companies don’t target the security industry, their unique solutions could apply there. Jevans believes strongly in the promise of blockchain for video security. “It’s totally the right way to do it,” he says, sounding surprised that it hadn’t occurred to him earlier.
The problem is scale. Submitting every second of a video to a blockchain to be signed and timestamped is expensive on high-energy blockchains like bitcoin. However, Jevans brings up some alternatives. The EOS blockchain has a “pretty fast rate,” he says. “I’ve heard of [private] blockchains that can do one million transactions per second, some that can do two or three million,” he adds, though he was “not at liberty to say” which ones.
“You could have a million cameras connected up to such a blockchain that could commit a video every second,” says Jevans. “With ten or twenty of those, you could cover the entire world’s security cameras.”