Video is now ubiquitous in our daily
lives. Law enforcement, in particular, relies on it for investigations in
addition to footage from witnesses and other sources. Reliance on video has
opened up an unforeseen threat, “deepfakes.” The integrity of video footage has
now come into question as the proliferation of software allowing video
manipulation increases. Fortunately there is a project on the horizon that
takes aim in curbing “deepfakes” via cryptographic authentication.
Amber
Authenticate, a tool developed to run silently in the background during video
capture, can generate hashes at user defined intervals. Hashes are then
recorded on a public blockchain which can be used to authenticate the same
footage snippet through the algorithm at a later date. Dissimilar hashes tip
the user off to the possibility of tampering.
Best practices should be used when setting hashing intervals
of course. Too long of an interval could result in quick, undetectable
tampering, while too short of an interval may be cause system constraints and
can be overkill. Situations that would benefit from short intervals would law
enforcement body camera footage while storefront CCTV camera recordings would
best be suited for long intervals.
Shamir Allibhai, Amber CEO, expresses that systemic risk
with police body cameras inherent to many manufacturers and models and the
threat of deepfakes can make it almost impossible to detect a fake once entered
as evidence. Currently, detection is always a step behind, which Amber
Authenticate aims to bring in-step with efforts to manipulate video.
Human rights activists, free speech advocates, and law
enforcement oversight committees see great potential in a tool like Amber
Authenticate in exposing cover-ups of abuse. The governments can also benefit
from a video integrity tool as well. Allibhai presented his video cryptography
solution to Department of Defense and Department of Homeland Security
representatives. Also in attendance were vendors like Factom who are also
working on a similar video authentication tool.
Built on popular open-source blockchain platform Ethereum,
Amber Authenticate includes a web-based Graphical User Interface. The interface
provides feedback, in the form of a green frame, when the video is authentic,
and a red frame when a mismatched hash has been identified. Amber Authenticate
also displays an “audit trail” listing original file creation, uploaded,
hashed, and submitted to the blockchain.
Amber Authenticate creator Shamir Allibhai is confident
manufacturers will want to license his software for use in CCTV and body camera
applications. This comes after his research consultant Josh Mitchell was able
to find vulnerabilities in five different models of popular body cameras.
Mitchell was able to prove compatibility of Amber Authenticate with some of
those popular brands.
Video authentication tools for body cameras in particular is
a long time coming for this platform. Civil liberty union policy analyst Jay
Stanley says that Amber Authenticate will have to be evaluated against industry
standards for tools of this caliber. Stanley is hopeful that other products
like Amber become the available standard and perhaps provide confidence in
evidence back to communities.
References
Newman, L. H. (2019, February 11). A New Tool
Protect Videos From Deepfakes and Tampering. Retrieved from Wired.com:
https://www.wired.com/story/amber-authenticate-video-validation-blockchain-tampering-deepfakes/
No comments:
Post a Comment