How error-prone data systems are reshaping your rights and risks

EPIC filed a brief with the Supreme Court to stop Arizona from using a federal database to cancel voter registrations. The state uses the Systematic Alien Verification for Entitlements (SAVE) system to check every voter against a list of noncitizens. If the system finds a match, the state assumes the person is not a citizen and removes them from the voter rolls immediately, even in the 90 days before an election when such removals are usually banned. EPIC

The core issue is that the SAVE system was not originally built to verify citizenship for voting purposes. It was designed for other government benefits, and it has a known problem with accuracy. When a system is used for a purpose it was not designed for, mistakes become much more likely. This means that a simple data match can trigger a serious legal consequence without a human ever checking if the match is actually correct. EPIC

Arizona and the federal government argue that this bulk matching process does not violate the federal quiet period rules. They believe that because the state is using a federal database, it is allowed to bypass the protections that normally stop mass removals of voters right before an election. This legal argument is currently being tested in the Supreme Court case Republican National Committee v. Mi Familia Vota. EPIC

This affects anyone in Arizona who is registered to vote, but it hits U.S. citizens the hardest. The Department of Homeland Security has admitted that the SAVE system has data accuracy problems that make it unreliable for confirming citizenship. Because of this, U.S. citizens have been mistakenly identified as noncitizens and had their registrations canceled in other states that use the same system. EPIC

The risk is not just about being wrong on a list; it is about the difficulty of fixing the mistake. Once a registration is canceled, re-registering is a long and burdensome process. For many people, especially those who are elderly or have limited access to government offices, this delay can effectively prevent them from voting in the next election. EPIC

This is not an isolated incident. EPIC has been fighting against the misuse of the SAVE system for years. They led a coalition to criticize changes to the system and are currently a plaintiff in a separate lawsuit challenging the expansion of SAVE into a national citizenship database. This shows that the problem is systemic and has been growing for some time. EPIC

This case is a clear example of how automated decision-making can override human rights without proper oversight. When a computer says you are not a citizen, the system treats that as a fact, even if it is wrong. This is the same logic that drives many of the privacy concerns you have about AI: algorithms make decisions that affect your life, and it is very hard to appeal those decisions if the system is opaque. EPIC

The Supreme Court’s decision here will set a precedent for how states can use federal data to enforce their own laws. If the Court rules in favor of Arizona, other states may feel emboldened to use similar databases for other types of eligibility checks, such as for driver’s licenses or public benefits. This could lead to a broader trend of using error-prone data to restrict access to services. EPIC

For you, this means that your data is being used in ways you did not agree to. You likely provided your information for a specific purpose, such as registering to vote, but that data is now being cross-referenced with other government systems to make decisions about your rights. This is a key reason why data minimization and purpose limitation are so important in privacy law. EPIC

This story connects to a larger trend of governments using data systems to enforce compliance in ways that are difficult for individuals to challenge. The SAVE system is just one example, but it highlights a fundamental problem: when data is collected for one purpose and then used for another, the safeguards that were put in place for the original purpose often do not apply. This is a common theme in privacy debates, from social media algorithms to credit scoring. EPIC

The fact that this is happening in the 90 days before an election makes it especially urgent. The quiet period was designed to prevent exactly this kind of last-minute disruption. If the Court allows this exception, it could undermine the stability of the voter rolls in other states as well. This is not just a legal technicality; it is a direct threat to the integrity of the electoral process. EPIC

As you think about your own data, consider how it might be used in ways you did not expect. Your voter registration, your health records, your financial history—all of these are being stored in systems that can be accessed and used by different agencies for different purposes. Understanding how these systems work is the first step to protecting your rights in an increasingly data-driven world. EPIC

• A data breach at the German political party AfD exposed 345,000 email addresses, with the data later published online with AI-generated summaries. Have I Been Pwned
• In Dayton, Minnesota, Flock reinstalled two AI surveillance cameras after the city explicitly ordered them removed, citing privacy concerns and previous misuse. Tom's Hardware
• A leak of DHS records exposed the contact information and internal details of 41,000 employees, including undercover agents, from the agency's Okta login system. WIRED

1. Check your voter registration status. If you live in Arizona or a state that uses similar data-matching systems, verify that your registration is still active. You can do this through your state’s election website or by contacting your local election office. This takes less than five minutes and ensures you are not unexpectedly removed from the rolls. EPIC

1. Review your data footprint. Take ten minutes to check which services you have signed up for and what data they collect. You can use tools like Have I Been Pwned to see if your email has been exposed in breaches. This helps you understand where your data is going and what risks you are taking. Have I Been Pwned