Category: Privacy & Security
-
Decoding China’s Ambitious Generative AI Regulations
By Sihao Huang and Justin Curl On April 11th, 2023, China’s top internet regulator proposed new rules for generative AI. The draft builds on previous regulations on deep synthesis technology, which contained detailed provisions on user identity registration, the creation of a database of undesirable inputs, and even the generation of “special objects and scenes”…
-
Unrecoverable Election Screwup in Williamson County TX
In the November 2020 election in Williamson County, Texas, flawed e-pollbook software resulted in voters inadvertently voting for candidates and questions not from their own districts but from others in the same county. These voters were deprived of the opportunity to vote for candidates they were entitled to vote for—and their votes were wrongly counted…
-
Next Steps for Mercer County Following Voting-Machine Failure
Hand-marked optical-scan paper ballots are the most secure form of voting: with any other method, if the computerized voting machines are hacked, there’s no trustworthy paper trail from which we can determine the true outcome of the election, based on the choices that voters actually indicated. Even those voting methods that appear to have a…
-
The anomaly of cheap complexity
Why are our computer systems so complex and so insecure? For years I’ve been trying to explain my understanding of this question. Here’s one explanation–which happens to be in the context of voting computers, but it’s a general phenomenon about all our computers: There are many layers between the application software that implements an electoral…
-
Toward Trustworthy Machine Learning: An Example in Defending against Adversarial Patch Attacks (2)
By Chong Xiang and Prateek Mittal In our previous post, we discussed adversarial patch attacks and presented our first defense algorithm PatchGuard. The PatchGuard framework (small receptive field + secure aggregation) has become the most popular defense strategy over the past year, subsuming a long list of defense instances (Clipped BagNet, De-randomized Smoothing, BagCert, Randomized…
-
Toward Trustworthy Machine Learning: An Example in Defending against Adversarial Patch Attacks
By Chong Xiang and Prateek Mittal Thanks to the stunning advancement of Machine Learning (ML) technologies, ML models are increasingly being used in critical societal contexts — such as in the courtroom, where judges look to ML models to determine whether a defendant is a flight risk, and in autonomous driving, where driverless vehicles are…
-
Switzerland’s E-voting: The Threat Model
Part 5 of a 5-part series starting here Switzerland commissioned independent expert reviews of the E-voting system built by Swiss Post. One of those experts concluded, “as imperfect as the current system might be when judged against a nonexistent ideal, the current system generally appears to achieve its stated goals, under the corresponding assumptions…
-
What the Assessments Say About the Swiss E-voting System
(Part 4 of a 5-part series starting here) In 2021 the Swiss government commissioned several in-depth technical studies of the Swiss Post E-voting system, by independent experts from academia and private consulting firms. They sought to assess, does the protocol as documented guarantee the security called for by Swiss law (the “ordinance on electronic voting”,…
-
How the Swiss Post E-voting system addresses client-side vulnerabilities
(Part 3 of a 5-part series starting here) In Part 1, I described how Switzerland decided to assess the security and accuracy of its e-voting system. Swiss Post is the “vendor” developing the system, the Swiss cantons are the “customer” deploying it in their elections, and the Swiss Parliament and Federal Chancellery are the “regulators,” …