Two recent statements from the surveillance company—one addressing Illinois privacy violations and another defending the company's national surveillance network—reveal a troubling pattern: when confronted by evidence of widespread abuse, Flock Safety has blamed users, downplayed harms, and doubled down on the very systems that enabled the violations in the first place. Flock's aggressive public relations campaign to salvage its reputation comes as no surprise. Last month, we described how investigative reporting from 404 Media revealed that a sheriff's office in Texas searched data from more than 83,000 automated license plate reader (ALPR) cameras to track down a woman suspected of self-managing an abortion. (A scenario that may have been avoided, it's worth noting, had Flock taken action when they were first warned about this threat three years ago). Flock calls the reporting on the Texas sheriff's office "purposefully misleading," claiming the woman was searched for as a missing person at her family's request rather than for her abortion. But that ignores the core issue: this officer used a nationwide surveillance dragnet (again: over 83,000 cameras) to track someone down, and used her suspected healthcare decisions as a reason to do so. Framing this as concern for her safety plays directly into anti-abortion narratives that depict abortion as dangerous and traumatic in order to justify increased policing, criminalization, control—and, ultimately, surveillance. Flock Safety has blamed users, downplayed harms, and doubled down on the very systems that enabled the violations in the first place. As if that weren't enough, the company has also come under fire for how its ALPR network data is being actively used to assist in mass deportation. Despite U.S. Immigration and Customs Enforcement (ICE) having no formal agreement with Flock Safety, public records revealed "more than 4,000 nation and statewide lookups by local and state police done either at the behest of the federal government or as an 'informal' favor to federal law enforcement, or with a potential immigration focus." The network audit data analyzed by 404 exposed an informal data-sharing environment that creates an end-run around direct oversight and accountability measures: federal agencies can access the surveillance network through local partnerships without the transparency and legal constraints that would apply to direct federal contracts. Flock Safety is adamant this is "not Flock's decision," and by implication, not their fault. Instead, the responsibility lies with each individual local law enforcement agency. In the same breath, they’re adamant that data sharing is essential, loudly claiming credit when the technology is involved in cross-jurisdictional investigations—but failing to show the same attitude when that data-sharing ecosystem is used to terrorize abortion seekers or immigrants.  Flock Safety: The Surveillance Social Network In growing from a 2017 startup to a $7.5 billion company "serving over 5,000 communities, Flock allowed individual agencies wide berth to set and regulate their own policies. In effect, this approach offered cheap surveillance technology with minimal restrictions, leaving major decisions and actions in the hands of law enforcement while the company scaled rapidly. And they have no intention of slowing down. Just this week, Flock launched its Business Network, facilitating unregulated data sharing amongst its private sector security clients. "For years, our law enforcement customers have used the power of a shared network to identify threats, connect cases, and reduce crime. Now, we're extending that same network effect to the private sector," Flock Safety's CEO announced.  Flock Safety wooing law enforcement officers at the 2023 International Chiefs of Police Conference. The company is building out a new mass surveillance network using the exact template that ended with the company having to retrain thousands of officers in Illinois on how not to break state law—the same template that made it easy for officers to do so in the first place. Flock's continued integration of disparate surveillance networks across the public and private spheres—despite the harms that have already occurred—is owed in part to the one thing that it's gotten really good at over the past couple of years: facilitating a surveillance social network.  Employing marketing phrases like "collaboration" and "force multiplier," Flock encourages as much sharing as possible, going as far as to claim that network effects can significantly improve case closure rates. They cultivate a sense of shared community and purpose among users so they opt into good faith sharing relationships with other law enforcement agencies across the country. But it's precisely that social layer that creates uncontrollable risk. The possibility of human workarounds at every level undermines any technical safeguards Flock may claim. Search term blocking relies on officers accurately labeling search intent—a system easily defeated by entering vague reasons like "investigation" or incorrect justifications, made either intentionally or not. And, of course, words like "investigation" or "missing person" can mean virtually anything, offering no value to meaningful oversight of how and for what the system is being used. Moving forward, sheriff's offices looking to avoid negative press can surveil abortion seekers or immigrants with ease, so long as they use vague and unsuspecting reasons.  The same can be said for case number requirements, which depend on manual entry. This can easily be circumvented by reusing legitimate case numbers for unauthorized searches. Audit logs only track inputs, not contextual legitimacy. Flock's proposed AI-driven audit alerts, something that may be able to flag suspicious activity after searches (and harm) have already occurred, relies on local agencies to self-monitor misuse—despite their demonstrated inability to do so. Flock operates as a single point of failure that can compromise—and has compromised—the privacy of millions of Americans simultaneously. And, of course, even the most restrictive department policy may not be enough. Austin, Texas, implemented one of the most restrictive ALPR programs in the country, and the program still failed: the city's own audit revealed systematic compliance failures that rendered its guardrails meaningless. The company's continued appeal to "local policies" means nothing when Flock's data-sharing network does not account for how law enforcement policies, regulations, and accountability vary by jurisdiction. You may have a good relationship with your local police, who solicit your input on what their policy looks like; you don't have that same relationship with hundreds or thousands of other agencies with whom they share their data. So if an officer on the other side of the country violates your privacy, it’d be difficult to hold them accountable.  ALPR surveillance systems are inherently vulnerable to both technical exploitation and human manipulation. These vulnerabilities are not theoretical—they represent real pathways for bad actors to access vast databases containing millions of Americans' location data. When surveillance databases are breached, the consequences extend far beyond typical data theft—this information can be used to harass, stalk, or even extort. The intimate details of people's daily routines, their associations, and their political activities may become available to anyone with malicious intent. Flock operates as a single point of failure that can compromise—and has compromised—the privacy of millions of Americans simultaneously. Don't Stop de-Flocking Rather than addressing legitimate concerns about privacy, security, and constitutional rights, Flock has only promised updates that fall short of meaningful reforms. These software tweaks and feature rollouts cannot assuage the fear engendered by the massive surveillance system it has built and continues to expand. A typical specimen of Flock Safety's automated license plate readers. Flock's insistence that what's happening with abortion criminalization and immigration enforcement has nothing to do with them—that these are just red-state problems or the fault of rogue officers—is concerning. Flock designed the network that is being used, and the public should hold them accountable for failing to build in protections from abuse that cannot be easily circumvented. Thankfully, that's exactly what's happening: cities like Austin, San Marcos,  Denver, Norfolk, and San Diego are pushing back. And it's not nearly as hard a choice as Flock would have you believe: Austinites are weighing the benefits of a surveillance system that generates a hit less than 0.02% of the time against the possibility that scanning 75 million license plates will result in an abortion seeker being tracked down by police, or an immigrant being flagged by ICE in a so-called "sanctuary city." These are not hypotheticals. It is already happening. Given how pervasive, sprawling, and ungovernable ALPR sharing networks have become, the only feature update we can truly rely on to protect people's rights and safety is no network at all. And we applaud the communities taking decisive action to dismantle its surveillance infrastructure. Follow their lead: don't stop de-flocking.

Today’s decision in Free Speech Coalition v. Paxton is a direct blow to the free speech rights of adults. The Court ruled that “no person—adult or child—has a First Amendment right to access speech that is obscene to minors without first submitting proof of age.” This ruling allows states to enact onerous age-verification rules that will block adults from accessing lawful speech, curtail their ability to be anonymous, and jeopardize their data security and privacy. These are real and immense burdens on adults, and the Court was wrong to ignore them in upholding Texas’ law.   Importantly, the Court's reasoning applies only to age-verification rules for certain sexual material, and not to age limits in general. We will continue to fight against age restrictions on online access more broadly, such as on social media and specific online features.   Still, the decision has immense consequences for internet users in Texas and in other states that have enacted similar laws. The Texas law forces adults to submit personal information over the internet to access entire websites that hold some amount of sexual material, not just pages or portions of sites that contain specific sexual materials. Many sites that cannot reasonably implement age verification measures for reasons such as cost or technical requirements will likely block users living in Texas and other states with similar laws wholesale.   Importantly, the Court's reasoning applies only to age-verification rules for certain sexual material, and not to age limits in general.  Many users will not be comfortable sharing private information to access sites that do implement age verification, for reasons of privacy or concern for data breaches. Many others do not have a driver’s license or photo ID to complete the age verification process. This decision will, ultimately, deter adult users from speaking and accessing lawful content, and will endanger the privacy of those who choose to go forward with verification.  What the Court Said Today  In the 6-3 decision, the Court ruled that Texas’ HB 1181 is constitutional. This law requires websites that Texas decides are composed of “one-third” or more of “sexual material harmful to minors” to confirm the age of users by collecting age-verifying personal information from all visitors—even to access the other two-thirds of material that is not adult content.    In 1997, the Supreme Court struck down a federal online age-verification law in Reno v. American Civil Liberties Union. In that case the court ruled that many elements of the Communications Decency Act violated the First Amendment, including part of the law making it a crime for anyone to engage in online speech that is "indecent" or "patently offensive" if the speech could be viewed by a minor. Like HB 1181, that law would have resulted in many users being unable to view constitutionally protected speech, as many websites would have had to implement age verification, while others would have been forced to shut down.   In Reno and in subsequent cases, the Supreme Court ruled that laws that burden adults’ access to lawful speech are subjected to the highest level of review under the First Amendment, known as strict scrutiny. This level of scrutiny requires a law to be very narrowly tailored or the least speech-restrictive means available to the government.   That all changed with the Supreme Court’s decision today.   The Court now says that laws that burden adults’ access to sexual materials that are obscene to minors are subject to less-searching First Amendment review, known as intermediate scrutiny. And under that lower standard, the Texas law does not violate the First Amendment. The Court did not have to respond to arguments that there are less speech-restrictive ways of reaching the same goal—for example, encouraging parents to install content-filtering software on their children’s devices. The court reached this decision by incorrectly assuming that online age verification is functionally equivalent to flashing an ID at a brick-and-mortar store. As we explained in our amicus brief, this ignores the many ways in which verifying age online is significantly more burdensome and invasive than doing so in person. As we and many others have previously explained, unlike with in-person age-checks, the only viable way for a website to comply with an age verification requirement is to require all users to upload and submit—not just momentarily display—a data-rich government-issued ID or other document with personal identifying information.   This leads to a host of serious anonymity, privacy, and security concerns—all of which the majority failed to address. A person who submits identifying information online can never be sure if websites will keep that information or how that information might be used or disclosed. This leaves users highly vulnerable to data breaches and other security harms. Age verification also undermines anonymous internet browsing, even though courts have consistently ruled that anonymity is an aspect of the freedom of speech protected by the First Amendment.     This Supreme Court broke a fundamental agreement between internet users and the state that has existed since its inception The Court sidestepped its previous online age verification decisions by claiming the internet has changed too much to follow the precedent from Reno that requires these laws to survive strict scrutiny. Writing for the minority, Justice Kagan disagreed with the premise that the internet has changed: “the majority’s claim—again mistaken—that the internet has changed too much to follow our precedents’ lead.”    But the majority argues that past precedent does not account for the dramatic expansion of the internet since the 1990s, which has led to easier and greater internet access and larger amounts of content available to teens online. The majority’s opinion entirely fails to address the obvious corollary: the internet’s expansion also has benefited adults. Age verification requirements now affect exponentially more adults than they did in the 1990s and burden vastly more constitutionally protected online speech. The majority's argument actually demonstrates that the burdens on adult speech have grown dramatically larger because of technological changes, yet the Court bizarrely interprets this expansion as justification for weaker constitutional protection.  What It Means Going Forward  This Supreme Court broke a fundamental agreement between internet users and the state that has existed since its inception: the government will not stand in the way of people accessing First Amendment-protected material. There is no question that multiple states will now introduce similar laws to Texas. Two dozen already have, though they are not all in effect. At least three of those states have no limit on the percentage of material required before the law applies—a sweeping restriction on every site that contains any material that the state believes the law includes. These laws will force U.S.-based adult websites to implement age-verification or block users in those states, as many have in the past when similar laws were in effect.   Rather than submit to verification, research has found that people will choose a variety of other paths: using VPNs to indicate that they are outside of the state, accessing similar sites that don’t comply with the law, often because the site is operating in a different country. While many users will simply not access the content as a result, others may accept the risk, at their peril.    We expect some states to push the envelope in terms of what content they consider “harmful to minors,” and to expand the type of websites that are covered by these laws, either through updated language or threats of litigation. Even if these attacks are struck down, operators of sites that involve sexual content of any type may be under threat, especially if that information is politically divisive. We worry that the point of some of these laws will be to deter queer folks and others from accessing lawful speech and finding community online by requiring them to identify themselves. We will continue to fight to protect against the disclosure of this critical information and for people to maintain their anonymity.  EFF Will Continue to Fight for All Users’ Free Expression and Privacy  That said, the ruling does not give states or Congress the green light to impose age-verification regulations on the broader internet. The majority’s decision rests on the fact that minors do not have a First Amendment right to access sexual material that would be obscene. In short, adults have a First Amendment right to access those sexual materials, while minors do not. Although it was wrong, the majority’s opinion ruled that because Texas is blocking minors from speech they have no constitutional right to access, the age-verification requirement only incidentally burdens adult’s First Amendment rights.   But the same rationale does not apply to general-audience sites and services, including social media. Minors and adults have coextensive rights to both speak and access the speech of other users on these sites because the vast majority of the speech is not sexual materials that would be obscene to minors. Lawmakers should be careful not to interpret this ruling to mean that broader restrictions on minors’ First Amendment rights, like those included in the Kids Online Safety Act, would be deemed constitutional.   Free Speech Coalition v. Paxton will have an effect on nearly every U.S. adult internet user for the foreseeable future. It marks a worrying shift in the ways that governments can restrict access to speech online. But that only means we must work harder than ever to protect privacy, security, and free speech as central tenets of the internet.  

A Georgia court has decided that private non-profit Atlanta Police Foundation (APF) must comply with public records requests under the Georgia Open Records Act for some of its functions on behalf of the Atlanta Police Department. This is a major win for transparency in the state.   The lawsuit was brought last year by the Atlanta Community Press Collective (ACPC) and Electronic Frontier Alliance member Lucy Parsons Labs (LPL). It concerns the APF’s refusal to disclose records about its role as the leaser and manager of the site of so-called Cop City, the Atlanta Public Safety Training Center at the heart of a years-long battle that pitted local social and environmental movements against the APF. We’ve previously written about how APF and similar groups fund police surveillance technology, and how the Atlanta Police Department spied on the social media of activists opposed to Cop City.   This is a big win for transparency and for local communities who want to maintain their right to know what public agencies are doing.  Police Foundations often provide resources to police departments that help them avoid public oversight, and the Atlanta Police Foundation leads the way with its maintenance of the Loudermilk Video Intergration Center and its role in Cop City, which will be used by public agencies including the Atlanta and other police departments.  ACPC and LPL were represented by attorneys Joy Ramsingh, Luke Andrews, and Samantha Hamilton who had won the release of some materials this past December. The plaintiffs had earlier been represented by the University of Georgia School of Law First Amendment Clinic.   The win comes at just the right time. Last Summer, the Georgia Supreme Court ruled that private contractors working for public entities are subject to open records laws. The Georgia state legislature then passed a bill to make it harder to file public records requests against private entities. With this month’s ruling, there is still time for the Atlanta Police Foundation to appeal the decision, but failing that, they will have to begin to comply with public records requests by the beginning of July.   We hope that this will help ensure transparency and accountability when government agencies farm out public functions to private entities, so that local activists and journalists will be able to uncover materials that should be available to the general public. 

Things are speeding up in generative AI legal cases, with two judicial opinions just out on an issue that will shape the future of generative AI: whether training gen-AI models on copyrighted works is fair use. One gets it spot on; the other, not so much, but fortunately in a way that future courts can and should discount. The core question in both cases was whether using copyrighted works to train Large Language Models (LLMs) used in AI chatbots is a lawful fair use. Under the US Copyright Act, answering that question requires courts to consider: whether the use was transformative; the nature of the works (Are they more creative than factual? Long since published?) how much of the original was used; and the harm to the market for the original work. In both cases, the judges focused on factors (1) and (4). The right approach In Bartz v. Anthropic, three authors sued Anthropic for using their books to train its Claude chatbot. In his order deciding parts of the case, Judge William Alsup confirmed what EFF has said for years: fair use protects the use of copyrighted works for training because, among other things, training gen-AI is “transformative—spectacularly so” and any alleged harm to the market for the original is pure speculation.Just as copying books or images to create search engines is fair, the court held, copying books to create a new, “transformative” LLM and related technologies is also protected: [U]sing copyrighted works to train LLMs to generate new text was quintessentially transformative. Like any reader aspiring to be a writer, Anthropic’s LLMs trained upon works not to race ahead and replicate or supplant them—but to turn a hard corner and create something different. If this training process reasonably required making copies within the LLM or otherwise, those copies were engaged in a transformative use. Importantly, Bartz rejected the copyright holders’ attempts to claim that any model capable of generating new written material that might compete with existing works by emulating their “sweeping themes, “substantive points,” or “grammar, composition, and style” was an infringement machine. As the court rightly recognized, building gen-AI models that create new works is beyond “anything that any copyright owner rightly could expect to control.”  There’s a lot more to like about the Bartz ruling, but just as we were digesting it Kadrey v. Meta Platforms came out. Sadly, this decision bungles the fair use analysis. A fumble on fair use Kadrey is another suit by authors against the developer of an AI model, in this case Meta’s ‘Llama’ chatbot. The authors in Kadrey asked the court to rule that fair use did not apply. Much of the Kadrey ruling by Judge Vince Chhabria is dicta—meaning, the opinion spends many paragraphs on what it thinks could justify ruling in favor of the author plaintiffs, if only they had managed to present different facts (rather than pure speculation). The court then rules in Meta’s favor because the plaintiffs only offered speculation.  But it makes a number of errors along the way to the right outcome. At the top, the ruling broadly proclaims that training AI without buying a license to use each and every piece of copyrighted training material will be “illegal” in “most cases.” The court asserted that fair use usually won’t apply to AI training uses even though training is a “highly transformative” process, because of hypothetical “market dilution” scenarios where competition from AI-generated works could reduce the value of the books used to train the AI model.. That theory, in turn, depends on three mistaken premises. First, that the most important factor for determining fair use is whether the use might cause market harm. That’s not correct. Since its seminal 1994 opinion in Cambell v Acuff-Rose, the Supreme Court has been very clear that no single factor controls the fair use analysis. Second, that an AI developer would typically seek to train a model entirely on a certain type of work, and then use that model to generate new works in the exact same genre, which would then compete with the works on which it was trained, such that the market for the original works is harmed. As the Kadrey ruling notes, there was no evidence that Llama was intended to to, or does, anything like that, nor will most LLMs for the exact reasons discussed in Bartz. Third, as a matter of law, copyright doesn't prevent “market dilution” unless the new works are otherwise infringing. In fact, the whole purpose of copyright is to be an engine for new expression. If that new expression competes with existing works, that’s a feature, not a bug. Gen-AI is spurring the kind of tech panics we’ve seen before; then, as now, thoughtful fair use opinions helped ensure that copyright law served innovation and creativity. Gen-AI does raise a host of other serious concerns about fair labor practices and misinformation, but copyright wasn’t designed to address those problems. Trying to force copyright law to play those roles only hurts important and legal uses of this technology. In keeping with that tradition, courts deciding fair use in other AI copyright cases should look to Bartz, not Kadrey.

This week, EFF joined EDRi and nearly 50 civil society organizations urging the European Commission’s President Ursula von der Leyen, Executive Vice President Henna Virkunnen, and Commissioners Michael McGrath and Hadja Lahbib to take immediate action and defend human rights in Hungary. The European Commission has a responsibility to protect EU fundamental rights, including the rights of LGBTQ+ individuals in Hungary and across the Union With Budapest Pride just two days away, Hungary has criminalized Pride marches and is planning to deploy real-time facial recognition technology to identify those participating in the event. This is a flagrant violation of fundamental rights, particularly the rights to free expression and assembly. On April 15, a new amendment package went into effect in Hungary which authorizes the use of real-time facial recognition to identify protesters at ‘banned protests’ like LGBTQ+ events, and includes harsh penalties like excessive fines and imprisonment. This is prohibited by the EU Artificial Intelligence (AI) Act, which does not permit the use of real-time face recognition for these purposes. This came on the back of members of Hungary’s Parliament rushing through three amendments in March to ban and criminalize Pride marches and their organizers, and permit the use of real-time facial recognition technologies for the identification of protestors. These amendments were passed without public consultation and are in express violation of the EU AI Act and Charter of Fundamental Rights. In response, civil society organizations urged the European Commission to put interim measures in place to rectify the violation of fundamental rights and values. The Commission is yet to respond—a real cause of concern. This is an attack on LGBTQ+ individuals, as well as an attack on the rights of all people in Hungary. The letter urges the European Commission to take the following actions: Open an infringement procedure against any new violations of EU law, in particular the violation of Article 5 of the AI Act Adopt interim measures on ongoing infringement against Hungary’s 2021 anti LGBT law which is used as a legal basis for the ban on LGBTQIA+ related public assemblies, including Budapest Pride. There's no question that, when EU law is at stake, the European Commission has a responsibility to protect EU fundamental rights, including the rights of LGBTQ+ individuals in Hungary and across the Union. This includes ensuring that those organizing and marching at Pride in Budapest are safe and able to peacefully assemble and protest. If the EU Commission does not urgently act to ensure these rights, it risks hollowing out the values that the EU is built from. Read our full letter to the Commission here.