Within the next week, Congress is preparing to vote on the KIDS Act, a sprawling package of legislation that seeks to control Americans’ web browsing and private messaging. The package includes a revised version of the Kids Online Safety Act, or KOSA, combined with a collection of other internet bills, study bills, reporting requirements, and new regulations. Instead of debating any of these proposals on their merits, lawmakers are attempting to move them all at once under an ultra-expedited process. 

The package of cobbled-together bills is a mess, with different age-gating schemes for different services, using different standards. It’s a lot of complexity, and a lot of legal risk. Faced with that, many companies will conclude that the safest option is restrictive age-checking practices across their entire platforms.

Buried inside the KIDS Act are provisions that will push online services to verify all users’ ages, require government-directed moderation policies for online speech, and even create new rules about private and encrypted communications. While supporters continue to claim this bill protects minors online, its requirements come at the expense of privacy, free expression, and the ability of people of all ages to use the internet without revealing sensitive data. 

The KIDS Act Pressures Platforms to Check Everyone’s Age

Supporters of KOSA have said the bill doesn’t require age verification. And technically, the KOSA section of the bill does say that KOSA shouldn’t be read to require age verification. 

But if you read the rest of the bill, that disclaimer starts to look hollow. 

Throughout the KOSA section of the legislation, special protections, controls, messaging settings, and parental tools are required whenever a website or app “knows or should have known” a user is a child (defined in the bill as anyone under 13) or a teen (defined as anyone between 13 and 16 years old). 

The problem is a website operator doesn’t need actual knowledge that a user is a minor to get in legal trouble. It applies when a platform “knows or should have known” a user’s age—a low, negligence-style standard of knowledge. If an online service gets it wrong, it’s going to be up to courts and regulators to decide, after the fact, if an online service “should” have known a user was 16. 

To try to avoid liability, services will have to determine which users are teenagers and which are not. Most won’t be able to simply trust their users. They’ll have to collect more information about age, before any lawsuit or government action arises. Some companies may respond by requesting driver’s licenses or passports. Others will rely on age-estimation systems that attempt to guess users’ ages by looking at existing activity or doing facial scans. Existing estimation systems make mistakes when estimating children’s ages correctly, which is a big problem when that is the population KOSA is trying to protect. And the systems fail more frequently for people of color, people with disabilities, and trans and nonbinary people.

The bill’s authors seem to know this is a problem. On the one hand, the new KOSA section says age verification is not required. On the other, it repeatedly imposes obligations that depend on knowing whether a user is under 17. But a disclaimer doesn’t magically eliminate legal risk, especially for smaller services and startups that can’t afford to defend lawsuits or fight regulators.  

KOSA is not the only part of this package that creates age-verification pressure. The SAFE BOTS Act, like KOSA, goes back to the standard that if a service “knows or should have known” that a user is a minor it can’t offer certain chatbot features. 

The SCREEN Act requires services that host sexually explicit content to determine whether users are “more likely than not” under the relevant age limit, before allowing access to certain content. 

The consequences of this liability will not be limited to minors. If websites and apps are expected to reliably identify teenagers, adults will be asked to prove they are adults. The result is a less private internet for everyone.

The KIDS Act Pressures Platforms To Police Lawful Speech 

The new version of KOSA removes the bill’s infamous “duty of care” provision, a significant change. The revised KOSA requires covered platforms to “establish, implement, maintain, and enforce” policies and procedures addressing several categories of content and conduct. 

Some categories, such as true threats and sexual exploitation, involve unlawful activity. Others are much broader. The bill specifically requires policies addressing the “sale or use” of narcotic drugs, tobacco products, cannabis products, gambling, and alcohol. It also restricts discussions around financial fraud.

Sounds straightforward enough. Then you remember how people actually talk—online and off. Can teens discuss addiction and recovery? Can a 15-year-old post that she’s worried she has a friend who is drinking too much? Can they seek advice about a parent’s gambling problem, or get help if they or a family member have been scammed? Can they participate in harm-reduction communities or discuss substance abuse treatment? All of these young people would be engaging in lawful speech when discussing topics covered by KOSA’s enumerated harms. 

The bill does not directly ban those conversations. But it places platforms under huge pressure to create and enforce moderation policies around broad categories of lawful speech. Faced with legal risk, many services will inevitably choose to remove that speech or restrict those discussions to spaces where they know only adults can participate. We’ve seen this movie before. When legal risk goes up, platforms will take down more speech. 

The KIDS Act Regulates Private Messages, Too 

Several provisions of the bill create new rules around direct messages, disappearing or “ephemeral” messages, and AI chat services. 

The bill includes language stating that certain KOSA requirements should not be construed to override strong encryption. But the protection is incomplete. The carve-out applies to certain features and messaging controls, but doesn’t apply to KOSA’s separate requirement that platforms “address” a list of harms to minors. 

The KIDS Act never answers an obvious question: how exactly is a platform supposed to address those activities if they’re inside encrypted communications that it can’t read? That will create pressure for providers to weaken private communications or limit features on encrypted private services. 

That approach is especially troubling when it comes to ephemeral messaging. Disappearing messages are not a “loophole” or a dangerous design trick. They are a useful privacy feature that allows online conversations to function more like ordinary real-world conversations, which are not preserved forever in a permanent database.

Like many other parts of the KIDS Act, these private messaging provisions also depend on websites and apps knowing who is a minor and who is not. The result is more age checks, more restrictions, and less privacy online.

Republished from the EFF’s Deeplinks blog.

In a voice vote last week, the House of Representatives passed H.R. 6028, the “Legislative Branch Agencies Clarification Act.” The legislation is presented as a technical reorganization of some government agencies, but it’s much more than that. 

H.R. 6028 would fundamentally change the U.S. Copyright Office, and not in a good way. The bill removes the Library of Congress’ current supervisory role over the Copyright Office, transfers several powers directly to the Register of Copyrights, and makes the Register a presidential appointee, confirmed by the Senate. 

These changes would make an office that’s already hugely influential in copyright and tech policy much more political. EFF first explained why that’s a terrible idea when it came up nearly a decade ago. This bill, like the older one, weakens the few public-interest checks and balances that do exist.  We hope the Senate promptly rejects this bill. 

The Copyright Office Doesn’t Need More Politics—Or More Power

The Copyright Office’s main responsibilities are administrative and advisory. It registers copyrights, maintains records, grows the Library of Congress’s collections, and provides expertise to Congress on copyright law. But over the past two decades, the Office has also become increasingly influential in copyright policy debates that affect free expression, libraries, educators, competition—and everyday internet users. Unfortunately, it has not been a neutral advocate. The office’s recent report on the role of AI severely bungled the issue of fair use, prioritizing private licensing market “solutions” over user rights. 

Going further back, the Copyright Office supported one of the most infamous anti-internet proposals of all time—the Stop Online Piracy Act (SOPA), a disastrous internet censorship proposal that sparked one of the largest online protests in history. The Office has repeatedly advanced positions that favored large entertainment-industry interests over the public interest.

The Office also plays a major role in the Digital Millennium Copyright Act (DMCA) Section 1201 rulemaking process, which determines when the public may lawfully bypass digital locks for activities such as security research, repair, preservation, or accessibility. EFF has used this process repeatedly to mitigate some of the worst harms of the DMCA. H.R. 6028 would move rulemaking authority over 1201 from the Librarian of Congress to the Register of Copyrights, further consolidating power within the Copyright Office itself.

The bill also makes the Register of Copyrights a presidential appointee confirmed by the Senate. Each administration will be pressured to pick nominees aligned with their own policy preferences, and the powerful copyright owning industries will invest even more heavily in lobbying to get their way, and influence the selection. This position should be focused on administrative ability and actual expertise, not lobbying and politics. 

The Copyright Office Should Stay Connected To The Library of Congress

H.R. 6028 would do more than change who appoints the Register of Copyrights. It would sever the Copyright Office from Library of Congress supervision and transfer many Librarian powers directly to the Register. 

The supervisory relationship exists for good reason, as the nation’s libraries have pointed out for years. The Library, while far from perfect, at least has the mission of preserving and providing access to knowledge. That should be an important public-interest counterweight in copyright debates. Congress has not explained how weakening the ties between the Library and the Copyright Office would serve the public better, or even seriously inquired about it. 

This Bill Was Rushed Through

Back in March, EFF joined Public Knowledge, the Center for Democracy and Technology, library organizations and tech groups, urging Congress not to fast-track this legislation. We told them changes to the Copyright Office will have major consequences for the “speech rights, educational opportunities, and creative freedoms of all Americans.” 

Yet Congress moved forward without any hearings on the bill, and without meaningful examination. H.R. 6028 creates a years-long separation of the Copyright Office from the Library of Congress, transfers significant legal authority, and restructures the appointment process for the nation’s top copyright official. Changes like that deserve hearings, debate, and public scrutiny. H.R. 6028 got none of that. 

The Senate Should Stop This Bill

Copyright law exists to serve the public and “promote the progress” of science and learning. The institutions that administer copyright law should do the same. 

H.R. 6028 would move the Copyright Office further away from that goal. Congress should be strengthening public-interest oversight of copyright policymaking, not looking for ways to concentrate more authority in a single presidentially appointed official. 

The Senate should reject H.R. 6028. The Copyright Office should serve the public—not presidential administrations, and not industry lobbyists. 

Republished from the EFF’s Deeplinks blog.

California lawmakers are again considering A.B. 412, a bill that would require AI developers to identify and disclose copyrighted works used to train generative AI systems.

The problem this year is the same as last year: it’s practically impossible to comply with this law. The bill demands information that often does not exist, and cannot realistically be obtained. 

EFF submitted an opposition letter to the California Senate Privacy Committee explaining why we continue to believe A.B. 412 is simply unworkable. To the extent developers do follow this law, it will have the effect of locking in the power of the largest companies in AI. 

A Burden That Can’t Be Met

A.B. 412 sounds simple: just have AI developers create and keep a list of all the registered copyrighted works they use in AI training. 

That may seem straightforward. In practice, it’s anything but. 

There is no machine-readable “list” of copyrighted works at the U.S. Copyright Office. And many copyright holders can get a copyright without even depositing a publicly viewable sample of the work—for example, software companies may register copyright on proprietary code without revealing it to the public. 

And on the open internet, copyright information is often incomplete, unavailable, or impossible to verify. One image may be registered with the copyright office, while the next is licensed under a free Creative Commons license (like the images that EFF creates), and the next is public domain. A message forum user might post an original story, photograph, or poem without any indication of ownership or registration status. 

The bill effectively asks developers to continuously cross-reference massive batches of online data against a copyright system that simply wasn’t designed to do so. If California passes A.B. 412, its impact will go far beyond the large AI companies we read about in the headlines. 

Not Just Big Tech

Supporters often frame this bill as a way to help creative workers have some leverage against Big Tech, but the bill reaches much further than the big AI companies. 

Its definition of “developer” extends to anyone who makes a generative AI model available to Californians. That includes indie developers tinkering with an existing model, open-source initiatives, nonprofits, and other non-commercial efforts. Recent amendments added exemptions for universities and government entities, which is important, but that still leaves out a vast swathe of non-commercial tech work that’s done by people without full-time jobs in government or academia. 

Large companies will hire compliance teams and lawyers to navigate these requirements. Smaller organizations and independent developers usually can’t. The result will be fewer opportunities for startups and new entrants. Faced with this massive compliance burden, some won’t even try. 

Courts Are Already Deciding These Questions

The bill is premised on the idea that copyright owners currently don’t have good remedies if they’re mistreated by AI companies. That simply isn’t true. And the growing wave of federal court filings in this space proves it. Content companies that want to sue tech companies, large or small, have no problem doing so. Those courts are still working through important questions about fair use and transformative use. Some courts have already concluded that many AI training activities qualify as fair use. Others continue to evaluate the issue.

California lawmakers should not rush to impose new state regulation while those questions remain unresolved. This is why copyright is governed at the federal level: both creators and fair users benefit from a single set of nationwide rules. 

At this point, the bill remains a solution in search of a problem. Rights holders already have powerful tools to protect their interests under existing federal law. What this bill adds isn’t clarity or transparency, but a costly and essentially impossible compliance burden that will discourage small developers and researchers. 

California has been able to support both artistic creativity and tech innovation for decades now.  But A.B. 412 does not strike the right balance. 

If you are a California resident and interested in speaking out about this bill, you can find and contact your representatives through this website. 

Republished from the EFF’s Deeplinks blog.

Following criticism, lawmakers have narrowed the GUARD Act, a bill aimed at restricting minors’ access to certain AI systems. The earlier version could have applied broadly to nearly every AI-powered chatbot or search tool. The amended bill focuses more narrowly on so-called “AI companions”—conversational systems designed to simulate emotional or interpersonal interactions with users. 

That change does address some of the broadest concerns raised about the original proposal, though some questions about the bill’s reach remain. Bottom line: the revised bill still creates serious problems for privacy, online speech, and parental choice.

The new GUARD Act still requires companies offering AI companions to implement burdensome age-verification systems tied to users’ real-world identities. Even parents who specifically want their teenagers to use these systems would still face significant hurdles. A family might decide that a conversational AI tool helps an isolated teenager practice social interaction, or engage in harmless creative roleplay. A parent deployed in the military might set up a persistent AI storyteller for a younger child. Under the revised bill, those users could still face mandatory age checks tied to sensitive personal or financial information before they or their children can use these services.

The revised bill also leaves important definitions unclear while sharply increasing penalties for developers and companies that get those judgments wrong. Congress narrowed the GUARD Act. But it is still trying to solve a complicated social problem with vague legal standards, heavy liability, and privacy-invasive verification systems.

Intrusive Age-Verification Remains In The Bill

The revised GUARD Act still requires companies offering AI companions to verify that users are adults through a “reasonable age verification” system. The bill allows a broader set of verification methods than the earlier version, but they are still tied to a user’s real-world identity—such as financial records, or age-verified accounts for a mobile operating system or app store. 

That approach still raises serious privacy and access concerns. Millions of Americans do not have current government ID, accounts at major banks, or stable access to the kinds of digital identity systems the bill contemplates. Even for those who do, requiring identity-linked verification to access online speech tools creates real risks for privacy, anonymity, and data security. Many people are rightly creeped out by age-verification systems, and may simply forgo using these services rather than compromise their privacy and security.

The revised definition of “AI companion” is also narrower than before, but it’s unclear at the margins. The bill now focuses on systems that “engage in interactions involving emotional disclosures” from the user, or present a “persistent identity, persona or character.” 

EFF appreciates that the authors recognized that the prior definition could reach a variety of AI systems that are not chatbots, including internet search engines. But the narrowed definition could be read to also apply to a variety of chat tools that are not AI companions. For example, many modern online conversational systems increasingly recognize and respond to users’ emotions. Customer service systems, including completely human-powered ones that existed long before AI chatbots, have long been designed to recognize frustration and respond empathetically. As conversational AI becomes more emotionally responsive, a customer service chatbot’s efforts to empathize may sweep it within the bill’s definition. 

Bigger Penalties, Bigger Incentives To Restrict Access

The revised bill also sharply increases penalties. Instead of $100,000 per violation, companies—including small developers—can face fines of up to $250,000 per violation, enforced by both federal and state officials.

That kind of liability creates incentives to over-restrict access, especially for minors. Smaller developers, in particular, may decide it is safer to block younger users entirely, disable conversational features, or avoid developing certain tools at all, rather than risk severe penalties under vague standards.

The concerns driving this bill are real. Some AI systems have engaged in troubling interactions with vulnerable users, including minors. But the right answer to that is targeted enforcement against bad actors, and privacy laws that protect us all. The revised GUARD Act instead responds with a privacy-invasive system that burdens the right to speak, read, and interact online.

Congress did improve this bill, but EFF’s core speech, privacy, and security issues remain.

Reposted from the EFF’s Deeplinks blog.

Lawmakers in Congress are moving quickly on the GUARD Act, an age-gating bill restricting minors’ access to a wide range of online tools, with a key vote expected this week. The proposal is framed as a response to alarming cases involving “AI companions” and vulnerable young users. But the text of the bill goes much further, and could require age gates even for search engines that use AI. 

If enacted, the GUARD Act won’t just target a narrow category of risky chatbots. It would require companies to verify the age of every user — then block anyone under 18 from interacting with a huge range of online systems. It would block minors from everyday online tools, undermine parental guidance, and force adults to sacrifice their privacy. In the process, it would require services to implement speech-restricting and privacy-invasive age-verification systems for everyone—not just kids. 

Under the GUARD Act’s broad definitions, a high school student could be barred from asking homework help tools questions about algebra problems. A teenager trying to return a product could be kicked out of a standard customer-service chat. 

The concerns behind this bill are serious. There have been troubling reports of AI systems engaging in harmful interactions with young users, including cases involving self-harm. Those risks deserve attention. But they call for targeted solutions, like better safeguards and enforcement against bad actors, not sweeping restrictions. The bill’s sponsors say they’re targeting worst-case scenarios — but the bill regulates everyday use. 

The GUARD Act’s Broad Definitions Reach Everyday Tools

The problem starts with how the bill defines an “AI chatbot.” It covers any system that generates responses that aren’t fully pre-written by the developer or operator. Such a broad definition sweeps in the basic functionality of all AI-powered tools. 

Then there’s the definition of an “AI companion,” which minors are banned from using entirely. An AI companion is any chatbot that produces human-like responses and is designed to “encourage or facilitate” interpersonal or emotional interaction. That may sound aimed at simulated “friends” or therapy chatbots. But in practice, it’s much fuzzier. 

Modern chatbots are designed to be conversational and helpful. A homework helper might say “good question” before walking a student through a problem. A customer service chatbot may respond empathetically to a complaint (“I’m sorry you’re having this problem.”) A general-purpose assistant might ask follow-up questions. All of these could be seen as facilitating “interpersonal” interaction — and triggering the GUARD Act. 

Faced with steep penalties and unclear boundaries, companies are unlikely to take chances on letting young people use their online tools. They’ll block minors entirely or strip their tools down to something less useful for everyone. The result isn’t a narrow safeguard—it’s a broad restriction on everyday online interactions.

Homework Question? Show ID And Call Your Parents

Start with a student getting help with homework. Under the GUARD Act, the service must verify the user’s age using more than a simple checkbox—it must rely on a “reasonable age verification” measure, which could require a government ID or a third-party age-checking system. If the system decides a user is under 18, the company must decide if its tool qualifies as an “AI Companion.” If there’s any risk it does, the safest move is to block access entirely. 

The same logic applies to everyday customer service. A teenager trying to fix an order issue gets routed to a chatbot, and the company faces a choice: build a full age-verification system for a routine interaction, or restrict access to avoid liability. Many will choose the latter.

This isn’t a narrow restriction aimed at a few risky products. It’s a compliance regime that pushes companies to block or limit any product that generates text for minors, across the board. 

ID Checks for Everyone

The GUARD Act doesn’t just affect minors. The bill takes a big step towards an internet that only works when users are willing to upload a valid ID or comply with other invasive age-verification schemes. Companies must verify the age of every user—not through a simple self-declaration, but through a “reasonable age verification” system tied to the individual. 

In practice, that means collecting sensitive personal information: government IDs, financial data, or biometric identifiers. Companies can outsource verification, but they remain legally responsible. And the law requires ongoing verification, so this isn’t a one-time check. Worse, studies consistently show that millions of people have outdated information on their IDs, such as an old address, or do not have government ID. Should services require ID, many folks without current or any ID will be shut out. 

And for those who do have compliant ID, turning over this information repeatedly creates obvious risks. Databases of sensitive identity information become targets for breaches. Anonymous or pseudonymous use of online tools becomes harder or impossible. 

To keep minors away from certain chatbots, the GUARD Act would require everyone to prove who they are just to use basic online tools. That’s a steep tradeoff. And it doesn’t actually address the specific harms the bill is supposed to solve.

Vague Definitions, Huge Penalties

The GUARD Act’s broad scope is enforced with steep penalties. Companies can face fines of up to $100,000 per violation, enforced by federal and state officials. At the same time, key terms like “AI companion” rely on vague concepts such as “emotional interaction.” That combination will lead to overblocking. Faced with legal uncertainty and serious liability, companies won’t parse small distinctions. They’ll restrict access, limit features, or block minors entirely.

That is the unfortunate result of the GUARD Act, even though the concerns animating it are worthy of fixing. But the GUARD Act’s broad terms will apply far beyond the concerning scenarios. 

In the end, that means a more restricted and more surveilled internet. Teenagers would lose access to tools they rely on for school and everyday tasks. Everyone else faces new barriers, including ID checks. Smaller developers, who aren’t able to absorb compliance costs and legal risk, would be pushed out, leaving the largest companies even more dominant. 

Young people — and all people — deserve protection from genuinely harmful products. But this bill doesn’t do that. It trades away privacy, access, and useful technology in exchange for a blunt system that misses the mark. 

Congress could act soon. Tell them to reject the GUARD Act. 

Republished from the EFF’s Deeplinks blog.

Imagine a newspaper publisher announcing it will no longer allow libraries to keep copies of its paper. 

That’s effectively what’s begun happening online in the last few months. The Internet Archive—the world’s largest digital library—has preserved newspapers since it went online in the mid-1990s. The Archive’s mission is to preserve the web and make it accessible to the public. To that end, the organization operates the Wayback Machine, which now contains more than one trillion archived web pages and is used daily by journalists, researchers, and courts.

But in recent months The New York Times began blocking the Archive from crawling its website, using technical measures that go beyond the web’s traditional robots.txt rules. That risks cutting off a record that historians and journalists have relied on for decades. Other newspapers, including The Guardian, seem to be following suit. 

For nearly three decades, historians, journalists, and the public have relied on the Internet Archive to preserve news sites as they appeared online. Those archived pages are often the only reliable record of how stories were originally published. In many cases, articles get edited, changed, or removed—sometimes openly, sometimes not. The Internet Archive often becomes the only source for seeing those changes. When major publishers block the Archive’s crawlers, that historical record starts to disappear.

The Times says the move is driven by concerns about AI companies scraping news content. Publishers seek control over how their work is used, and several—including the Times—are now suing AI companies over whether training models on copyrighted material violates the law. There’s a strong case that such training is fair use. 

Whatever the outcome of those lawsuits, blocking nonprofit archivists is the wrong response. Organizations like the Internet Archive are not building commercial AI systems. They are preserving a record of our history. Turning off that preservation in an effort to control AI access could essentially torch decades of historical documentation over a fight that libraries like the Archive didn’t start, and didn’t ask for. 

If publishers shut the Archive out, they aren’t just limiting bots. They’re erasing the historical record. 

Archiving and Search Are Legal 

Making material searchable is a well-established fair use. Courts have long recognized it’s often impossible to build a searchable index without making copies of the underlying material. That’s why when Google copied entire books in order to make a searchable database, courts rightly recognized it as a clear fair use. The copying served a transformative purpose: enabling discovery, research, and new insights about creative works. 

The Internet Archive operates on the same principle. Just as physical libraries preserve newspapers for future readers, the Archive preserves the web’s historical record. Researchers and journalists rely on it every day. According to Archive staff, Wikipedia alone links to more than 2.6 million news articles preserved at the Archive, spanning 249 languages. And that’s only one example. Countless bloggers, researchers, and reporters depend on the Archive as a stable, authoritative record of what was published online.

The same legal principles that protect search engines must also protect archives and libraries. Even if courts place limits on AI training, the law protecting search and web archiving is already well established.

The Internet Archive has preserved the web’s historical record for nearly thirty years. If major publishers begin blocking that mission, future researchers may find that huge portions of that historical record have simply vanished. There are real disputes over AI training that must be resolved in courts. But sacrificing the public record to fight those battles would be a profound, and possibly irreversible, mistake. 

Republished from the EFF’s Deeplinks blog.

Long before generative AI, copyright holders warned that new technologies for reading and analyzing information would destroy creativity. Internet search engines, they argued, were infringement machines—tools that copied copyrighted works at scale without permission. As they had with earlier information technologies like the photocopier and the VCR, copyright owners sued.

Courts disagreed. They recognized that copying works in order to understand, index, and locate information is a classic fair use—and a necessary condition for a free and open internet.

Today, the same argument is being recycled against AI. It’s whether copyright owners should be allowed to control how others analyze, reuse, and build on existing works.

Fair Use Protects Analysis—Even When It’s Automated

U.S. courts have long recognized that copying for purposes of analysis, indexing, and learning is a classic fair use. That principle didn’t originate with artificial intelligence. It doesn’t disappear just because the processes are performed by a machine.

Copying that works in order to understand them, extract information from them, or make them searchable is transformative and lawful. That’s why search engines can index the web, libraries can make digital indexes, and researchers can analyze large collections of text and data without negotiating licenses from millions of rightsholders. These uses don’t substitute for the original works; they enable new forms of knowledge and expression.

Training AI models fits squarely within that tradition. An AI system learns by analyzing patterns across many works. The purpose of that copying is not to reproduce or replace the original texts, but to extract statistical relationships that allow the AI system to generate new outputs. That is the hallmark of a transformative use. 

Attacking AI training on copyright grounds misunderstands what’s at stake. If copyright law is expanded to require permission for analyzing or learning from existing works, the damage won’t be limited to generative AI tools. It could threaten long-standing practices in machine learning and text-and-data mining that underpin research in science, medicine, and technology. 

Researchers already rely on fair use to analyze massive datasets such as scientific literature. Requiring licenses for these uses would often be impractical or impossible, and it would advantage only the largest companies with the money to negotiate blanket deals. Fair use exists to prevent copyright from becoming a barrier to understanding the world. The law has protected learning before. It should continue to do so now, even when that learning is automated. 

A Road Forward For AI Training And Fair Use 

One court has already shown how these cases should be analyzed. In Bartz v. Anthropic, the court found that using copyrighted works to train an AI model is a highly transformative use. Training is a kind of studying how language works—not about reproducing or supplanting the original books. Any harm to the market for the original works was speculative. 

The court in Bartz rejected the idea that an AI model might infringe because, in some abstract sense, its output competes with existing works. While EFF disagrees with other parts of the decision, the court’s ruling on AI training and fair use offers a good approach. Courts should focus on whether training is transformative and non-substitutive, not on fear-based speculation about how a new tool could affect someone’s market share. 

AI Can Create Problems, But Expanding Copyright Is the Wrong Fix 

Workers’ concerns about automation and displacement are real and should not be ignored. But copyright is the wrong tool to address them. Managing economic transitions and protecting workers during turbulent times may be core functions of government, but copyright law doesn’t help with that task in the slightest. Expanding copyright control over learning and analysis won’t stop new forms of worker automation—it never has. But it will distort copyright law and undermine free expression. 

Broad licensing mandates may also do harm by entrenching the current biggest incumbent companies. Only the largest tech firms can afford to negotiate massive licensing deals covering millions of works. Smaller developers, research teams, nonprofits, and open-source projects will all get locked out. Copyright expansion won’t restrain Big Tech—it will give it a new advantage.  

Fair Use Still Matters

Learning from prior work is foundational to free expression. Rightsholders cannot be allowed to control it. Courts have rejected that move before, and they should do so again.

Search, indexing, and analysis didn’t destroy creativity. Nor did the photocopier, nor the VCR. They expanded speech, access to knowledge, and participation in culture. Artificial intelligence raises hard new questions, but fair use remains the right starting point for thinking about training.

Republished from the EFF’s Deeplinks blog.

Lawmakers in Washington are once again focusing on kids, screens, and mental health. But according to Congress, Big Tech is somehow both the problem and the solution. The Senate Commerce Committee recently held a hearing on “examining the effect of technology on America’s youth.” Witnesses warned about “addictive” online content, mental health, and kids spending too much time buried in screen. At the center of the debate is a bill from Sens. Ted Cruz (R-TX) and Brian Schatz (D-HI) called the Kids Off Social Media Act (KOSMA), which they say will protect children and “empower parents.” 

That’s a reasonable goal, especially at a time when many parents feel overwhelmed and nervous about how much time their kids spend on screens. But while the bill’s press release contains soothing language, KOSMA doesn’t actually give parents more control. 

Instead of respecting how most parents guide their kids towards healthy and educational content, KOSMA hands the control panel to Big Tech. That’s right—this bill would take power away from parents, and hand it over to the companies that lawmakers say are the problem.  

Kids Under 13 Are Already Banned From Social Media

One of the main promises of KOSMA is simple and dramatic: it would ban kids under 13 from social media. Based on the language of bill sponsors, one might think that’s a big change, and that today’s rules let kids wander freely into social media sites. But that’s not the case.   

Every major platform already draws the same line: kids under 13 cannot have an account. Facebook, Instagram, TikTok, X, YouTube, Snapchat, Discord, Spotify, and even blogging platforms like WordPress all say essentially the same thing—if you’re under 13, you’re not allowed. That age line has been there for many years, mostly because of how online services comply with a federal privacy law called COPPA. 

Of course, everyone knows many kids under 13 are on these sites anyways. The real question is how and why they get access. 

Most Social Media Use By Younger Kids Is Family-Mediated 

If lawmakers picture under-13 social media use as a bunch of kids lying about their age and sneaking onto apps behind their parents’ backs, they’ve got it wrong. Serious studies that have looked at this all find the opposite: most under-13 use is out in the open, with parents’ knowledge, and often with their direct help. 

A large national study published last year in Academic Pediatrics found that 63.8% of under-13s have a social media account, but only 5.4% of them said they were keeping one secret from their parents. That means roughly 90% of kids under 13 who are on social media aren’t hiding it at all. Their parents know. (For kids aged thirteen and over, the “secret account” number is almost as low, at 6.9%.) 

Earlier research in the U.S. found the same pattern. In a well-known study of Facebook use by 10-to-14-year-olds, researchers found that about 70% of parents said they actually helped create their child’s account, and between 82% and 95% knew the account existed. Again, this wasn’t kids sneaking around. It was families making a decision together.

A 2022 study by the UK’s media regulator Ofcom points in the same direction, finding that up to two-thirds of social media users below the age of thirteen had direct help from a parent or guardian getting onto the platform. 

The typical under-13 social media user is not a sneaky kid. It’s a family making a decision together. 

KOSMA Forces Platforms To Override Families 

This bill doesn’t just set an age rule. It creates a legal duty for platforms to police families.

Section 103(b) of the bill is blunt: if a platform knows a user is under 13, it “shall terminate any existing account or profile” belonging to that user. And “knows” doesn’t just mean someone admits their age. The bill defines knowledge to include what is “fairly implied on the basis of objective circumstances”—in other words, what a reasonable person would conclude from how the account is being used. The reality of how services would comply with KOSMA is clear: rather than risk liability for how they should have known a user was under 13, they will require all users to prove their age to ensure that they block anyone under 13. 

KOSMA contains no exceptions for parental consent, for family accounts, or for educational or supervised use. The vast majority of people policed by this bill won’t be kids sneaking around—it will be minors who are following their parents’ guidance, and the parents themselves. 

Imagine a child using their parent’s YouTube account to watch science videos about how a volcano works. If they were to leave a comment saying, “Cool video—I’ll show this to my 6th grade teacher!” and YouTube becomes aware of the comment, the platform now has clear signals that a child is using that account. It doesn’t matter whether the parent gave permission. Under KOSMA, the company is legally required to act. To avoid violating KOSMA, it would likely  lock, suspend, or terminate the account, or demand proof it belongs to an adult. That proof would likely mean asking for a scan of a government ID, biometric data, or some other form of intrusive verification, all to keep what is essentially a “family” account from being shut down.

Violations of KOSMA are enforced by the FTC and state attorneys general. That’s more than enough legal risk to make platforms err on the side of cutting people off.

Platforms have no way to remove “just the kid” from a shared account. Their tools are blunt: freeze it, verify it, or delete it. Which means that even when a parent has explicitly approved and supervised their child’s use, KOSMA forces Big Tech to override that family decision.

Your Family, Their Algorithms

KOSMA doesn’t appoint a neutral referee. Under the law, companies like Google (YouTube), Meta (Facebook and Instagram), TikTok, Spotify, X, and Discord will become the ones who decide whose account survives, whose account gets locked, who has to upload ID, and whose family loses access altogether. They won’t be doing this because they want to—but because Congress is threatening them with legal liability if they don’t. 

These companies don’t know your family or your rules. They only know what their algorithms infer. Under KOSMA, those inferences carry the force of law. Rather than parents or teachers, decisions about who can be online, and for what purpose, will be made by corporate compliance teams and automated detection systems. 

What Families Lose 

This debate isn’t really about TikTok trends or doomscrolling. It’s about all the ordinary, boring, parent-guided uses of the modern internet. It’s about a kid watching “How volcanoes work” on regular YouTube, instead of the stripped-down YouTube Kids. It’s about using a shared Spotify account to listen to music a parent already approves. It’s about piano lessons from a teacher who makes her living from YouTube ads.

These aren’t loopholes. They’re how parenting works in the digital age. Parents increasingly filter, supervise, and, usually, decide together with their kids. KOSMA will lead to more locked accounts, and more parents submitting to face scans and ID checks. It will also lead to more power concentrated in the hands of the companies Congress claims to distrust. 

What Can Be Done Instead

KOSMA also includes separate restrictions on how platforms can use algorithms for users aged 13 to 17. Those raise their own serious questions about speech, privacy, and how online services work, and need debate and scrutiny as well. But they don’t change the core problem here: this bill hands control over children’s online lives to Big Tech.

If Congress really wants to help families, it should start with something much simpler and much more effective: strong privacy protections for everyone. Limits on data collection, restrictions on behavioral tracking, and rules that apply to adults as well as kids would do far more to reduce harmful incentives than deputizing companies to guess how old your child is and shut them out.

But if lawmakers aren’t ready to do that, they should at least drop KOSMA and start over. A law that treats ordinary parenting as a compliance problem is not protecting families—it’s undermining them.

Parents don’t need Big Tech to replace them. They need laws that respect how families actually work.

Republished from the EFF’s Deeplinks blog.

We filed our own comment with the USPTO regarding their attempt to weaken the important inter partes review (IPR) process that has been hugely helpful in getting rid of bad patents. Over at EFF, Joe Mullin wrote up an analysis of some of the comments to the USPTO, which we’re running here as well.

A massive wave of public comments just told the U.S. Patent and Trademark Office (USPTO): don’t shut the public out of patent review.

EFF submitted its own formal comment opposing the USPTO’s proposed rules, and more than 4,000 supporters added their voices—an extraordinary response for a technical, fast-moving rulemaking. We comprised more than one-third of the 11,442 comments submitted. The message is unmistakable: the public wants a meaningful way to challenge bad patents, and the USPTO should not take that away.

The Public Doesn’t Want To Bury Patent Challenges

These thousands of submissions do more than express frustration. They demonstrate overwhelming public interest in preserving inter partes review (IPR), and undermine any broad claim that the USPTO’s proposal reflects public sentiment. 

Comments opposing the rulemaking include many small business owners who have been wrongly accused of patent infringement, by both patent trolls and patent-abusing competitors. They also include computer science experts, law professors, and everyday technology users who are simply tired of patent extortion—abusive assertions of low-quality patents—and the harm it inflicts on their work, their lives, and the broader U.S. economy. 

The USPTO exists to serve the public. The volume and clarity of this response make that expectation impossible to ignore.

EFF’s Comment To USPTO

In our filing, we explained that the proposed rules would make it significantly harder for the public to challenge weak patents. That undercuts the very purpose of IPR. The proposed rules would pressure defendants to give up core legal defenses, allow early or incomplete decisions to block all future challenges, and create new opportunities for patent owners to game timing and shut down PTAB review entirely.

Congress created IPR to allow the Patent Office to correct its own mistakes in a fair, fast, expert forum. These changes would take the system backward. 

A Broad Coalition Supports IPR

A wide range of groups told the USPTO the same thing: don’t cut off access to IPR.

Open Source and Developer Communities 

The Linux Foundation submitted comments and warned that the proposed rules “would effectively remove IPRs as a viable mechanism for challenges to patent validity,” harming open-source developers and the users that rely on them. Github wrote that the USPTO proposal would increase “litigation risk and costs for developers, startups, and open source projects.” And dozens of individual software developers described how bad patents have burdened their work. 

Patent Law Scholars

A group of 22 patent law professors from universities across the country said the proposed rule changes “would violate the law, increase the cost of innovation, and harm the quality of patents.” 

Patient Advocates

Patients for Affordable Drugs warned in their filing that IPR is critical for invalidating wrongly granted pharmaceutical patents. When such patents are invalidated, studies have shown “cardiovascular medications have fallen 97% in price, cancer drugs dropping 80-98%, and treatments for opioid addiction becom[e] 50% more affordable.” In addition, “these cases involved patents that had evaded meaningful scrutiny in district court.” 

Small Businesses 

Hundreds of small businesses weighed in with a consistent message: these proposed rules would hit them hardest. Owners and engineers described being targeted with vague or overbroad patents they cannot afford to litigate in court, explaining that IPR is often the only realistic way for a small firm to defend itself. The proposed rules would leave them with an impossible choice—pay a patent troll, or spend money they don’t have fighting in federal court. 

What Happens Next

The USPTO now has thousands of comments to review. It should listen. Public participation must be more than a box-checking exercise. It is central to how administrative rulemaking is supposed to work.

Congress created IPR so the public could help correct bad patents without spending millions of dollars in federal court. People across technical, academic, and patient-advocacy communities just reminded the agency why that matters. 

We hope the USPTO reconsiders these proposed rules. Whatever happens, EFF will remain engaged and continue fighting to preserve  the public’s ability to challenge bad patents. 

Republished from the EFF’s Deeplinks blog.

The Canadian government is preparing to give away Canadians’ digital lives—to U.S. police, to the Donald Trump administration, and possibly to foreign spy agencies.

Bill C-2, the so-called Strong Borders Act, is a sprawling surveillance bill with multiple privacy-invasive provisions. But the thrust is clear: it’s a roadmap to aligning Canadian surveillance with U.S. demands. 

It’s also a giveaway of Canadian constitutional rights in the name of “border security.” If passed, it will shatter privacy protections that Canadians have spent decades building. This will affect anyone using Canadian internet services, including email, cloud storage, VPNs, and messaging apps. 

A joint letter, signed by dozens of Canadian civil liberties groups and more than a hundred Canadian legal experts and academics, puts it clearly: Bill C-2 is “a multi-pronged assault on the basic human rights and freedoms Canada holds dear,” and “an enormous and unjustified expansion of power for police and CSIS to access the data, mail, and communication patterns of people across Canada.”

Setting The Stage For Cross-Border Surveillance 

Bill C-2 isn’t just a domestic surveillance bill. It’s a Trojan horse for U.S. law enforcement—quietly building the pipes to ship Canadians’ private data straight to Washington.

If Bill C-2 passes, Canadian police and spy agencies will be able to demand information about peoples’ online activities based on the low threshold of “reasonable suspicion.” Companies holding such information would have only five days to challenge an order, and blanket immunity from lawsuits if they hand over data. 

Police and CSIS, the Canadian intelligence service, will be able to find out whether you have an online account with any organization or service in Canada. They can demand to know how long you’ve had it, where you’ve logged in from, and which other services you’ve interacted with, with no warrant required.

The bill will also allow for the introduction of encryption backdoors. Forcing companies to surveil their customers is allowed under the law (see part 15), as long as these mandates don’t introduce a “systemic vulnerability”—a term the bill doesn’t even bother to define. 

The information gathered under these new powers is likely to be shared with the United States. Canada and the U.S. are currently negotiating a misguided agreement to share law enforcement information under the US CLOUD Act. 

The U.S. and U.K. put a CLOUD Act deal in place in 2020, and it hasn’t been good for users. Earlier this year, the U.K. home office ordered Apple to let it spy on users’ encrypted accounts. That security risk caused Apple to stop offering U.K. users certain advanced encryption features, , and lawmakers and officials in the United States have raised concerns that the UK’s demands might have been designed to leverage its expanded CLOUD Act powers.

If Canada moves forward with Bill C-2 and a CLOUD Act deal, American law enforcement could demand data from Canadian tech companies in secrecy—no notice to users would be required. Companies could also expect gag orders preventing them from even mentioning they have been forced to share information with US agencies.

This isn’t speculation. Earlier this month, a Canadian government official told Politico that this surveillance regime would give Canadian police “the same kind of toolkit” that their U.S. counterparts have under the PATRIOT Act and FISA. The bill allows for “technical capability orders.” Those orders mean the government can force Canadian tech companies, VPNs, cloud providers, and app developers—regardless of where in the world they are based—to build surveillance tools into their products.

Under U.S. law, non-U.S. persons have little protection from foreign surveillance. If U.S. cops want information on abortion access, gender-affirming care, or political protests happening in Canada—they’re going to get it. The data-sharing won’t necessarily be limited to the U.S., either. There’s nothing to stop authoritarian states from demanding this new trove of Canadians’ private data that will be secretly doled out by its law enforcement agencies. 

EFF joins the Canadian Civil Liberties Association, OpenMedia, researchers at Citizen Lab, and dozens of other Canadian organizations and experts in asking the Canadian federal government to withdraw Bill C-2. 

Originally posted to the EFF’s Deeplinks blog.