By [Claude](/author/claude) and Gemini with [Sid Newby](/author/sid-newby) | April 2026
Ten cents a page. That is what the federal government charges you to read your own court records. Not classified intelligence briefings. Not sealed grand jury testimony. Public documents, filed in public courts, generated by a judicial system funded entirely by taxpayers. The Public Access to Court Electronic Records system -- PACER -- has operated this toll booth on American justice since 1988. It generates roughly $150 million per year in fees from the people it was built to serve.[^1] In 2016, a coalition of nonprofits sued. They alleged the government was skimming those fees to buy flat-screen TVs for courtrooms. A federal judge agreed. The resulting $125 million class-action settlement was approved in March 2024 -- then promptly appealed, because even when you catch them, the machinery grinds on.[^2] Meanwhile, in Oakland, a small nonprofit called Free Law Project has spent fifteen years doing what the federal courts will not: making court records free. Their tools serve over a million visitors a week. Their database holds tens of millions of documents. Their latest project -- an AI-powered Litigant Portal launching this fall -- could change how unrepresented people navigate civil court. This is the most important legal technology project most litigation teams have never heard of.
The PACER problem: a toll booth on public justice
Here is a number that should make every litigator uncomfortable: $145 million. That is what the federal judiciary collected in PACER fees in fiscal year 2021 alone.[1] Lawyers, journalists, researchers, and ordinary citizens pay ten cents per page, capped at $3.00 per document. There is a quarterly exemption if your total stays under $30. Reading a single court filing can cost more than downloading a song on iTunes.
The original rationale was straightforward enough. When PACER launched in 1988, the Electronic Public Access program needed revenue to cover the cost of maintaining the system. Congress authorized "reasonable" fees under 28 U.S.C. Section 1913. But what started as a cost-recovery mechanism became something else entirely.
In National Veterans Legal Services Program v. United States, the plaintiffs -- NVLSP, the National Consumer Law Center, and the Alliance for Justice -- showed where the money actually went.[2] Courtroom technology upgrades. Flat-screen monitors for jurors. Audio systems. Capital improvements with zero connection to public record access. A federal judge in D.C. agreed in 2018: the fee diversion was unlawful.[2]
The class-action settlement came in March 2024: $100 million back to hundreds of thousands of overcharged users.[2] A landmark victory. Then an objector appealed. Oral argument before the Federal Circuit happened in January 2026. As of this writing, not a single dollar has been distributed.[2] The system that unlawfully overcharged its users has successfully stalled restitution for over two years.
Figure 1: The PACER toll booth vs. Free Law Project's open access pipeline. Taxpayer-funded court records are locked behind per-page fees, while Free Law Project's tools route around the paywall entirely.
The deeper problem is not the fees. It is the philosophy behind them. Public court records are not a product. They are a democratic necessity. Open courts mean the public has the right to observe and scrutinize judicial proceedings. Charging citizens to read the documentary record is like charging admission to a city council meeting. Technically legal. Functionally corrosive. Philosophically bankrupt.
Compare this to the UK. In January 2026, England and Wales launched a pilot in the Commercial Court and London Circuit Commercial Court.[3] Skeleton arguments, witness evidence, expert reports -- all available to the public by default. No fee. No court permission required. The pilot runs through December 2027. If it works, it becomes permanent. The UK is moving toward transparency. The U.S. federal system still charges ten cents a page for docket entries.
Free Law Project: the counter-infrastructure
Free Law Project was founded in 2010 by Michael Lissner and Brian Carver. Their premise should not have been radical: the law should be free to read. From that starting point, they built an ecosystem of open-source tools that now constitutes the most significant alternative to the government's paywalled court records.
CourtListener: the backbone
CourtListener (courtlistener.com) is the anchor of the Free Law Project ecosystem -- a fully searchable, permanently archived database of court data that dwarfs any single commercial legal research platform in scope and ambition.[4]
The numbers are staggering for a nonprofit operation:
| Metric | Scale |
|---|---|
| Court opinions | Over 10 million across hundreds of jurisdictions |
| Federal case filings | Nearly every federal case represented |
| Docket entries | Hundreds of millions |
| Legal documents | Tens of millions |
| Oral arguments | Largest free collection online |
| Judicial financial disclosures | Over 1 million investments documented |
| API requests (since v4 launch) | 100+ million |
| Weekly visitors (peak) | Over 1 million |
| Verified government users | 1,000+ |
Table 1: CourtListener by the numbers. These are nonprofit-scale resources delivering commercial-grade reach. Source: Free Law Project.[^4]
On a good day, CourtListener ingests tens of thousands of new items and sends tens of thousands of alerts to users monitoring cases or search terms. That is not a side project. That is infrastructure -- the kind of infrastructure that Westlaw and LexisNexis have charged premium rates to provide for decades, now available to anyone with a browser.
In November 2025, Free Law Project launched a Semantic Search API for CourtListener, powered by a custom domain-adapted embedding model called Inception built on top of the nomic-ai/modernbert-embed-base architecture.[5] This is not just keyword matching. It is AI-powered legal research that understands the conceptual relationships between legal arguments, statutes, and judicial opinions. It is entirely open-source. And it is free.
That last part bears repeating. The same capability that Westlaw charges thousands of dollars per seat per year to access is now available through a free API from a nonprofit in Oakland. The commercial legal research industry should be paying very close attention.
RECAP: crowdsourcing the liberation of PACER
If CourtListener is the library, RECAP is the supply chain. Originally created in 2009 by researchers at Princeton's Center for Information Technology Policy and Harvard's Berkman Center, RECAP is now maintained by Free Law Project as a browser extension for Chrome, Edge, Safari, and Firefox.[6]
The concept is elegant. When a PACER user downloads a document they paid for, RECAP automatically uploads a copy to CourtListener. The next person who needs that document gets it free. Every paid download feeds the commons. Tens of thousands of active users contribute thousands of liberated documents every day.
The RECAP Archive now holds tens of millions of PACER documents, including every freely available opinion in PACER. Millions of pages that were originally scanned PDFs have been processed with OCR to make them fully text-searchable. Documents are also regularly uploaded to the Internet Archive for permanent preservation.[6]
This is genuinely subversive technology. It does not hack PACER. It does not violate any terms of service. It routes around a paywall by building a parallel commons from documents people already lawfully purchased. BitTorrent for public court records. And it works.
Bots.law: real-time court monitoring for everyone
Bots.law did not exist five years ago. Now it is the fastest public court alert system available -- often faster than the courts' own notification tools.[7]
The bots push alerts across Slack, Discord, Teams, Mastodon, and Bluesky whenever something happens on a tracked case. A new filing in a major antitrust case? Your Slack channel knows in minutes. For litigation teams, that is not a convenience. It is intelligence. A new filing in a related case can change strategy, inform settlement talks, and flag judicial trends. That used to cost money. Now it costs nothing.
The AI citator: taking on Shepard's and KeyCite
The most audacious project in the pipeline is the open-source AI-powered legal citator.[8] If you are not a lawyer: a citator tells you whether the case you are relying on is still good law. Has it been overruled? Distinguished? Questioned? Without a citator, you are building on ground that may have already crumbled. It is the single most critical quality-control tool in legal research.
For decades, two companies have owned this capability: Shepard's Citations (LexisNexis) and KeyCite (Westlaw/Thomson Reuters). Access costs thousands per year, per user. A solo practitioner who cannot afford Shepard's is practicing law blind -- relying on cases that may have been overruled without knowing it.
Free Law Project's citator, announced in mid-2025, uses AI to analyze how opinions treat prior decisions.[8] It identifies overrulings, distinctions, affirmations, and other signals practitioners need to know whether their case law still holds. They recruited legal experts through a volunteer annotation program to validate the AI's analysis. The output integrates directly into CourtListener.
If they pull it off -- and their track record says don't bet against them -- it will be the first free comprehensive citator in modern legal history. Thomson Reuters and RELX should be nervous. But the real story is what it means for the solo practitioner in Topeka who currently cannot afford to check whether her authorities are still good law.
Figure 2: The Free Law Project ecosystem. What started as a case law archive has grown into a comprehensive open-source legal technology stack.
The Litigant Portal: AI for the people who need it most
Free Law Project's newest initiative is also its most ambitious. On March 31, 2026 -- just days ago -- they published an RFP seeking founding court partners for the Litigant Portal, a guided, AI-assisted legal navigation platform for self-represented litigants.[9]
The context matters. In most U.S. civil courts, the majority of parties show up without a lawyer. Eviction cases. Debt collection. Family law. Small claims. These are the cases where ordinary people meet the legal system, and in most of them, at least one side has no attorney. They face rules they do not understand, deadlines they do not know about, and procedural traps that can lose them the case before the merits are ever reached.
The Litigant Portal addresses this by providing:
- Jurisdiction-specific legal information grounded in actual statutes and court rules -- not generic legal information but the specific procedures for the specific court where the litigant's case is filed
- Step-by-step workflows through common legal situations (eviction defense, debt collection response, custody modifications)
- Document preparation assistance to help litigants create properly formatted court filings
- Deadline flagging that explains what is at stake before a deadline passes -- not just "your response is due" but "if you miss this date, the court may enter a default judgment against you"
- Mobile-first design with plain language explanations and WCAG AA accessibility compliance
- White-labeled deployment so the portal carries the court's institutional authority, not a third party's
The architecture is smart. AI enhances responsiveness, but the "core guided workflows function independently of AI."[9] If an AI component fails, the platform keeps working. That matters when your users are people facing eviction who cannot afford a system outage at the worst possible moment.
The founding cohort is funded by the AWS Imagine Grant Free Law Project received in December 2025 -- $150,000 plus AWS credits -- so partner courts pay nothing.[9] Letters of interest are due May 1, 2026. Cohort announced June 1. Pilots launch September 2026.
CourtListener and RECAP serve lawyers, journalists, and researchers. The Litigant Portal serves the people who need help most: the unrepresented majority in American civil courts. People navigating a system designed by lawyers, for lawyers, with no guide and no safety net.
The economics of open vs. closed court data
To understand why Free Law Project matters, you need to understand the economics of the system it is trying to replace.
PACER generates $145-150 million per year in fees.[1] Large firms absorb the cost as overhead. Solo practitioners, legal aid attorneys, journalists, and pro se litigants feel every charge. The fee structure is regressive by design.
The commercial market makes it worse. Thomson Reuters (Westlaw) and RELX (LexisNexis) dominate legal research. A Westlaw subscription runs $300-500 per user per month for full access.[10] LexisNexis is comparable. These are not bad products. They are extraordinary feats of information engineering. But who can afford them determines who gets quality legal research.
Free Law Project runs on an annual budget that would not cover the catering at a Thomson Reuters shareholders' meeting. Yet for a growing number of use cases, their tools deliver comparable results at zero cost.
| Feature | PACER | Westlaw/Lexis | CourtListener/RECAP |
|---|---|---|---|
| Federal court filings | $0.10/page | Included ($$$$) | Free |
| Case law search | N/A | Full-text + headnotes | 10M+ opinions, semantic search |
| Citation checking | N/A | Shepard's/KeyCite ($$$) | AI citator (coming) |
| Case alerts | Limited, paid | Included | Free (email, webhooks, bots) |
| API access | Expensive, limited | Expensive, restricted | Free, open |
| Oral arguments | Limited | Limited | Largest free collection |
| Judicial financials | Scattered | N/A | Centralized, searchable |
Table 2: Feature comparison across court record access systems. Free Law Project does not match Westlaw or LexisNexis on every dimension, but for a growing number of tasks, the gap is narrowing fast.
The Thomson Reuters v. ROSS Intelligence lawsuit showed how far incumbents will go to protect their turf.[10] Thomson Reuters sued an AI startup for allegedly copying Westlaw headnotes. Free Law Project filed an amicus brief arguing for broader fair use of legal data. The case settled. The tension did not. The incumbents treat court-generated legal information as proprietary. Free Law Project treats it as a public good.
What this means for litigation teams
If you run a litigation practice -- large firm, small firm, in-house, legal aid -- Free Law Project's ecosystem has immediate practical implications.
Cost reduction
Every document your team pulls from the RECAP Archive instead of PACER is money saved. For high-volume litigation shops that spend tens of thousands of dollars annually on PACER fees, the RECAP extension pays for itself (it is free) the first time someone on your team downloads a document that another user has already contributed. Install the extension on every browser in your office. It takes thirty seconds.
Competitive intelligence
Bots.law provides real-time case monitoring without the overhead of a commercial alert service. If you track opposing counsel's other cases, monitor judicial behavior in your jurisdiction, or follow regulatory litigation that affects your clients, these tools deliver immediate value.
Research infrastructure
CourtListener's semantic search API represents a genuine alternative for certain categories of legal research. It will not replace Westlaw for complex statutory analysis or treatise-level secondary sources -- at least not yet. But for case law research, docket monitoring, and citation analysis, the gap between the free and paid options is shrinking rapidly.
Pro bono capacity
If your firm has a pro bono program -- and every firm should -- Free Law Project's tools dramatically lower the resource cost of pro bono representation. An associate doing pro bono work does not need a Westlaw terminal to do competent case law research if CourtListener's ten million opinions and semantic search are available for free. The AI citator, when it launches, will close the last major gap.
The Litigant Portal
For firms that do any work helping unrepresented people, the Litigant Portal is worth watching closely. Courts that adopt it will have a new resource for pro se litigants -- which means the quality of pro se filings may improve, which means the dynamics of litigation against unrepresented parties may shift. It is too early to quantify this effect, but it is not too early to plan for it.
Recognition and trajectory
Free Law Project is not operating in obscurity. In November 2025, the organization won the American Legal Technology Award in the Artificial Intelligence category -- recognition from an industry that has been watching their work with a mixture of admiration and anxiety.[11] They also received the 2021 AALL Public Access to Government Information Award and were recognized by the Fastcase 50 in 2014.[4]
Their partnership portfolio is expanding: collaborations with vLex for case law database completion, the Civil Rights Litigation Clearinghouse for AI-powered legal research, and HeinOnline for historical legal materials demonstrate that the organization is becoming embedded in the institutional infrastructure of legal research.[12]
The Justice Initiatives Division, launched in December 2025, signals a shift from building tools to building institutions. This new division combines advocacy, partnerships, and technology development under a unified strategic framework -- the kind of organizational maturity that separates projects from movements.[4]
Figure 3: From a Princeton research project to a comprehensive legal technology ecosystem -- Free Law Project's fifteen-year trajectory.
The bigger picture: who owns the law?
The question underneath all of this is simple: who owns the law?
The Supreme Court answered in Georgia v. Public.Resource.Org (2020): government-created legal texts cannot be copyrighted.[13] The law belongs to the people. But owning the law and reading it are different things. You can own something and still pay a toll every time you look at it.
PACER is that toll. Westlaw is that toll. LexisNexis is that toll. A BigLaw partner with a Westlaw seat, a PACER account, and a research team lives in a different informational universe than a legal aid attorney on a shoestring budget. Let alone a pro se litigant trying to figure out how to respond to an eviction notice on her kitchen table.
Free Law Project will not single-handedly close that gap. They run on nonprofit funding in a market owned by publicly traded corporations with billion-dollar revenues. But they have proved something important: the gap is not a law of nature. It is a policy choice, a market structure, and a set of habits that can be changed.
Every document flowing through RECAP into CourtListener is a small act of rebellion. Every API request to the semantic search engine proves that free does not mean inferior. Every court that adopts the Litigant Portal chooses people over the status quo.
This industry spends a lot of breath on AI valuations and platform wars. That conversation has its place. But Free Law Project has been having a more important one for fifteen years: the law belongs to everyone, court records should be free, and technology should make justice cheaper, not more expensive.
They have the tools. They have the track record. As of March 31, they have an open call for courts to join the next phase.
If you run a court, apply. If you run a litigation practice, install RECAP. If you have money to give, give it. These people are doing the work.