About

What this is

Bad Faith Prosecution is an independent legal-research and citizen-journalism wiki by Danny at TheTatteredRose. Built to be corrected by readers, not just published at them. Any passage on this site can be highlighted; if you read something you know or believe is wrong, a small popup gives you the choice to comment publicly or send a private message. The people in similar situations are the ones who know what's missing. Your corrections matter more than the author's confidence.

The site examines the federal case against the Broadview Six (Kat Abughazaleh, Michael Rabbitt, Andre Martin, Catherine Sharp, Brian Straw, Joselyn Walsh) and makes the case for how they recover the legal fees the prosecution cost them. The durable route is the Hyde Amendment, a standing federal statute that lets a prevailing defendant recover fees when the government's position was vexatious, frivolous, or in bad faith. A second, contingent argument runs from the U.S. Department of Justice's own stated standards: by the rules of the DOJ's $1.776 billion Anti-Weaponization Fund, the defendants qualified. As of June 2026 that fund is blocked by a court order and the administration says it will not move forward with it, which is why the Hyde Amendment leads.

It is also the first in a series of wikis Danny is building. The point isn't only the Broadview SixKat Abughazaleh, Michael Rabbitt, Andre Martin, Catherine Sharp, Brian Straw, Joselyn Walsh case. The wiki is meant to be durable infrastructure that anyone in a similar position can use: the next defendant facing a politically-motivated federal prosecution, the next person who needs to know whether they can recover their legal fees, the next person who can't afford the kind of legal research that usually costs a retainer.

Who's making it

One person. Danny at TheTatteredRose. Computer science | sociology | advocacy.

This is the work of a citizen journalist working alone with AI. The honest term is “citizen journalism,” not “journalism” — the standard for journalism is a credential and an editorial process this site doesn't claim to have. The standard the site does claim is investigative-dossier exactness on every factual claim: receipts on the page, sources named, characterizations attributed to the court or named officials.

How it's made

This site is built with AI, but not the way most people are using AI right now. The entire codebase and every workflow that goes into making the site are open-source and posted publicly on GitHub. Anyone who wants to can see exactly how the AI was directed, paragraph by paragraph.

• The site uses MCPs (Model Context Protocols) and other tools that extend the out-of-the-box AI you can get from ChatGPT or any standard LLM. The biggest one: Midpage AI, which verifies every case-law citation against the actual court record before it appears on the site.
• Multiple AI agents coordinate with standardized handoffs and scoped knowledge of the project, so each agent has the right context at the right time. Not “one AI, one prompt, build a website.”
• Cross-model synthesis, including non-Western models, introduces methodological diversity beyond what any single Western AI would produce on its own.
• All of it is visible. The repository is public. Every commit is in the history. No “trust me, it's accurate” moves here. The receipts are on the page and in the git log.

The AI work is the starting line, not the finish. After a directed, source-checked first pass, the work is handed to humans, and only humans. Strict no-bot, no-auto-submissions policy. Every comment, every suggested change, every message gets manually reviewed by Danny before anything is published. Approved corrections get folded into the wiki over time, so it improves through real crowdsourcing the way “wisdom of the crowd” is supposed to work, minus the exploitationThe wisdom of the crowd gets sold all the time: prediction markets that take bets on whether a country goes to war, platforms that turn other people's suffering into a price. This wiki harnesses that collective work and aims it at the problem instead of skimming a profit from it..

Why use AI at all

Fair question. People we respect have a serious critique of AI: the energy it burns, the slop it floods the internet with, the jobs it puts at risk. None of that gets waved away here.

So here is what AI does on this site, and what it doesn't. It does not replace journalists or lawyers. The deep work comes from human experts who spent years earning it. The trouble is that the work is locked up, behind paywalls and retainers and the cost of a credential. If you're poor or working class and you're the one facing a federal prosecution, you can't afford the person who already knows whether the government's case was vexatious, frivolous, or in bad faith.

That's the gap this fills. AI here is a way to reach that expert knowledge, not to fake it. It puts the kind of legal research that normally costs a retainer within reach of the people most likely to need it and least able to pay for it, so they have a fighting chance to prevail and recover what the prosecution took from them. Every claim still gets checked against the court record and still carries a source you can open yourself. It is important to always keep in mind: this is citizen journalism, not journalism, and it is not legal advice. This project takes knowledge that is normally gatekept and puts it in the hands of the people who need it most, with the receipts shown and the risks named honestly instead of hand-waved.

On Audre Lorde

Using AI to do this work might look like a contradiction. There's a line you may know: “The master's tools will never dismantle the master's house.” Audre Lorde wrote it, the title of a 1979 talk later collected in Sister Outsider(1984). It isn't a contradiction, and Lorde is the reason it isn't.

Her warning was not that the oppressed may never pick up a tool. She spent her life on academic platforms; she did not tell anyone to stop writing. Her critique was about method, not machinery: the danger is repeating the so-called master's patterns — single-voice authority, gatekeeping, making the marginalized do the explaining. She was right then, and she is right now.

So we don't abandon the tool; we change how it's used. Walking away from AI because powerful companies built it would only lock the people who most need this knowledge further out of it. Retooling it, reshaping how it works and who it answers to, is the harder and better path, and it's the one this site takes: the whole pipeline is open-source, so the direction of the AI is visible instead of hidden; it runs on more than one company's model, including non-Western ones, so no single worldview is the default; every page is finished by humans, only humans. The substrate may be the so-called master's. The direction is ours. None of this is finished, and we don't claim it is. It's the move Lorde demanded, applied to a tool she did not live to see.

What it is NOT

This site is not legal advice and is not a substitute for a licensed attorney. Anyone considering a Hyde Amendment fee motion, a claim under the Anti-Weaponization Fund, or any other legal action, should consult counsel.

It is not a filed legal pleading. The arguments here are public-facing analysis, not something that can be submitted to a court.

It is not authorized by, endorsed by, or attributed to Ms. Abughazaleh, her campaign, her co-defendants Michael Rabbitt, Andre Martin, Catherine Sharp, Brian Straw, or Joselyn Walsh, their campaigns, or their legal counsel. No one connected to the case approved this work. The author has no contact relationship with the defendants or their representatives.

The receipts

Every factual claim on this site is drawn from public reporting and primary documents, verified as of the dates shown on each source citation. Characterizations of the prosecution are attributed to the findings of the United States District Court for the Northern District of Illinois and to the public statements of named officials. Where this site quotes a court transcript, the quotation is verbatim from the official record with a page-and-line citation.

The argument the site makes is one-sided by design. It states the strongest good-faith case for recovering the fees, and it states the obstacles candidly on their own pages. The durable claim is the Hyde Amendment, the standing fee-recovery statute; the contingent claim is that, by the Anti-Weaponization Fund's own stated rules, the defendants qualified. The site does not claim any defendant will win, should file, or that either route will pay. It argues the fee recovery in principle, the Hyde Amendment on its own terms and the Fund under the standards the DOJ itself stated.

The donate links

The site links to three legal-defense fundraisers: Kat + Andre's GoFundMe, Michael Rabbitt's GoFundMe, and Brian Straw's ActBlue. All three were organized by the defendants themselves. The defendants publicly asked supporters to give to any or all three. The links here honor that ask.

Each button opens that defendant's own verified canonical fundraiser URL. No payment is processed on this site. No share of any donation goes to this site or to Danny. The links do not imply that any defendant endorses this site.

The wiki way

This site is meant to be a wiki, which means it grows and improves through readers like you.

Any passage on the site is highlightable. If you read something you think (or know) is wrong, highlight it and a small popup will give you the choice to comment publicly or send a private message. Public comments are reviewed by Danny. Those that hold up under scrutiny get folded into the next iteration of the page. Private messages go directly to Danny's inbox.

Strict no-bot, no-auto-submissions policy. Manual review before any submission goes public. The point of the wiki form is that the people in similar situations are the ones who know what's missing and what's wrong. Their corrections matter more than any single author's confidence.