recentpopularlog in


« earlier   
Facial recognition is accurate - if you're a white guy
Study of facial recognition software conducted by Jyo Buolamwini at the MIT Media Lab finds that the darker the skin of the subject the more errors arise - up to 35% for darker-skinned women. At Georgetown Law School, researchers estimate that 117 million American adults are in face recognition networks used by law enforcement. African-Americans are disproportionately represented in mugshot databases.
biometrics  facial  recognition  race  fairness 
10 days ago by net.wars
Prevent Biometrics - Concussion Intelligence
revent’s patented technology was developed at Cleveland Clinic by top neurosurgeons and engineers over a period of 6 years with grants from the National Institute of Health and others, and is the only head impact monitor who’s accuracy has been validated in an IRB approved study published in the top peer reviewed Stapp Journal.
health  biometrics  concussion  wearable  sports  medical 
20 days ago by cyberchucktx
Biometrics in the News: MasterCard Plans for Transaction Implementation in 2019
Biometrics are on the forefront of the finance industry, as MasterCard announces their plans to implement these measures for in-store transactions starting in 2019.
biometrics  access-control  security 
21 days ago by Adventure_Web
Is India's Aadhaar system really "hack-proof"? Assessing a publicly observable security posture • Troy Hunt
<p>UIDAI is the Unique IDentification Authority of India and they run the Aadhaar project. Their statement echoes comments made around this latest incident that espouse the complete security of the system: "The Aadhaar data, including biometric information, is fully safe and secure".

Here's the issue I (and many others) have with these statements and I want to make it crystal clear:
Security is not a boolean proposition. It's not "secure" versus "insecure", "safe" versus "unsafe", rather it is a spectrum of controls that all contribute to an overall security posture. There is no "fully", there is no "completely"; every system - every single one - has weak points and a sufficiently well-equipped and determined adversary will find them.

It's the hubris of the UIDAI's statements which is the most worrying and it neglects so many of the highly sophisticated precedents that have come before the current situation. Precedents like Stuxnet, created by the US and Israeli governments to damage the Iranian nuclear program by targeting air-gapped centrifuges via 4 previously unknown "zero-day" flaws. That's almost a cliched example to pull out these days, the point is simply that where there is sufficient will and resources, any information system can be compromised.

But let's get back to that original tweet and the question therein: "Can you prove otherwise?" I certainly wouldn't want to be the person probing away at Aadhaar in an unauthorised fashion in order to prove otherwise (although make no mistake, many people are), but per the title of this post, there are many publicly observable things I can easily draw attention to. To be crystal clear, none of this is "hacking", it will merely involve looking at how the system responds to legitimate requests and observing the gap between what it does at present and what it ideally should do.</p>

Lengthy post. It's not certain that Aadhaar can be hacked, but one tends to think that where there's a will - and 1.2bn user records - there's a way.
aadhaar  biometrics  hacking 
6 weeks ago by charlesarthur

Copy this bookmark:

to read