State agencies in California are collecting and using more data now than they ever, and much of this data includes very personal information about California residents. This presents a challenge for agencies and the courts—how to make government-held data that’s indisputably of interest available to the public under the state’s public records laws while still protecting the privacy of Californians.

EFF filed an amicus brief today urging a state appeals court to reverse a San Francisco trial judge’s ruling that would impede and possibly preclude the public’s ability to access state-held data that includes private information on individuals—even if that data is anonymized or redacted to protect privacy.

The California Public Records Act (CPRA) has a strong presumption in favor of disclosure of state records. And the California state constitution recognizes that public access to government information is paramount in enabling people to oversee the government and ensure it’s acting in their best interest. But the state constitution also recognizes a strong privacy right, so public access to information must be carefully balanced with personal privacy.

To keep records secret, agencies must show that concealment, not transparency, best serves the public interest. This balancing test was at issue in a lawsuit brought by UCLA law professor Richard Sanders and the First Amendment Coalition (FAC), who are seeking access to information from the California Bar Association about the race, ethnicity, test scores, and other information of tens of thousands of people who took the state bar exam to become lawyers. The state bar refused to release the data to protect the confidentiality of test-takers, even though no names or personal identifying information would be disclosed. The case is Sander v. State Bar of California.

A trial court sided with the bar. The case eventually went all the way to the California Supreme Court, which correctly recognized the strong public interest in disclosing the data so the effect of law school admissions policies on exam performance could be studied. It’s “beyond dispute” that the public has the right to access the information, the court said in a unanimous decision, as long as the identity of individual test takers remained confidential. It sent the case back to the trial court to decide if and how much material could be released.

This is where things took a wrong turn. Sanders and FAC presented several possible protocols to protect bar exam takers’ privacy, including three complicated anonymization techniques, but the trial court ruled that, even under these protocols, the data couldn’t be released. The court improperly placed the burden on Sanders and FAC to show that there was absolutely no way anyone’s identity could be revealed—including if the anonymized data were combined with other obscure but publicly-available personal information. In doing so, the court failed to adhere to the CPRA’s balancing tests, which require the state bar to show that the public interest in protecting the privacy of bar takers—even after their data is stripped of identifying information—clearly outweighs the public interest in the data.

In a particularly dangerous finding, the court held the CPRA couldn’t require the state bar to apply anonymization protocols because that would constitute creating a “new record” from the existing data. However, the CPRA clearly requires agencies to produce as much public information as possible, even if that means using a “surgical scalpel” to separate information that’s exempt from disclosure under the CPRA from non-exempt information. Techniques for protecting exempt information while still releasing otherwise non-exempt government records that are of great interest to the public must evolve as the government’s means of collecting, compiling, and maintaining such records has evolved. Protocols that propose to anonymize data, such as those presented by Sander and FAC, represent one such technique. California courts should not avoid grappling with whether anonymization can protect privacy by dismissing it out of hand as the creation of a “new record.” 

The California’s Public Records Act is a vital check on government secrecy. With the explosive growth of government data, particularly law enforcement surveillance data, we can’t stand by while courts sidestep the task of evaluating anonymization protocols that will increasingly play a role in balancing public access rights under the CPRA and laws like it in other states. If upheld, the Sanders ruling could weaken the public’s ability to access other electronic records and government data that contains private identifying information. EFF has fought in court to gain access to license plate records indiscriminately collected on millions of drivers by Los Angeles law enforcement agencies. The California Supreme Court ruled that police can’t keep those records secret, paving the way for EFF to analyze how this huge surveillance program works. But the records could identify drivers, so the next step is to figure out how the data can be made public in a redacted or anonymized form to protect drivers’ privacy. We are watching the Sanders case closely, and hope the appeals court does the right thing: reverse the trial court’s findings, require it to fully address the proposed anonymization protocols, and properly apply the balancing tests under the CPRA.