There’s a certain symmetry to it. The man who secretly bankrolled the lawsuit that destroyed Gawker now wants to industrialise the process. No more ten-year court battles. No more millions in legal fees. Just $2,000 and an AI tribunal.
Objection AI launched on April 15 with backing from Peter Thiel, Balaji Srinivasan, and VC firms Social Impact Capital and Off Piste Capital. Its founder, Aron D’Souza — the same lawyer who orchestrated Thiel’s Gawker strategy — calls it “The AI Tribunal of Truth.” The pitch is seductive in its simplicity: anyone who feels wronged by a news story can pay to have its claims investigated and adjudicated by artificial intelligence.
But between the pitch and the reality sits a very uncomfortable question: who does this actually serve?
How Objection Works
The mechanics are straightforward enough. You file an objection against a specific, verifiable factual claim in a published story. You pay $2,000. Objection’s team — described as including former FBI, CIA, and NSA officials alongside investigative journalists — gathers evidence under what they call the “Empirical Journalism Standard” (EJ-1). The journalist is notified and can respond. Both sides submit evidence. Then a “jury” of large language models — OpenAI, Anthropic, xAI, Mistral, and Google’s models, coordinated by what Objection calls a “Judicial-Purpose Transformer” — evaluates the claim and issues a verdict: TRUE, FALSE, or INDETERMINABLE.
That verdict becomes a permanent public record, feeding into what Objection calls the Honor Index — a numerical score attached to a journalist’s name, marketed as a measure of their “integrity, accuracy, and track record.”
There’s also a companion feature called Fire Blanket, currently active on X, that posts real-time warnings about disputed claims while they’re still under investigation. Think of it as a digital scarlet letter that arrives before the trial.
D’Souza is blunt about the pedigree: “The Gawker litigation took ten years and millions of dollars. Objection industrialises this process.”
The Case For
Here’s the thing: the media does get things wrong. Sometimes catastrophically. The Rolling Stone campus rape story. The New York Times’ WMD reporting that helped launch a war. Brian Williams’ fabricated war stories. More recently, the flood of AI-generated slop passing as news across social platforms.
Journalism’s existing accountability mechanisms are slow, uneven, and largely self-policed. Corrections get buried at the bottom of articles. Ombudsmen are increasingly rare. Publications rarely face meaningful consequences for publishing inaccurate information, and the people harmed by those inaccuracies — particularly those without platforms of their own — have limited recourse.
D’Souza draws a parallel to X’s Community Notes, and it’s not entirely unfair. Community Notes has been one of the more successful experiments in crowdsourced fact-checking, precisely because it’s transparent, public, and doesn’t require a law degree to engage with. “It’s an attempt to fact-check,” D’Souza told TechCrunch. “The wisdom of the crowd plus the power of technology to create new methods of truth-telling.”
There is a genuine power asymmetry in media. Publications with millions of readers can damage reputations with a single paragraph, and the people they write about often have no practical way to respond in kind. A defamation lawsuit in the United States can cost hundreds of thousands of dollars and take years. Objection offers something that looks, at first glance, like a more democratic alternative.
UCLA First Amendment scholar Eugene Volokh frames it as part of the normal ecosystem of criticism that surrounds journalism — opposition research aimed at reporters instead of politicians. “All criticism creates a chilling effect,” he told TechCrunch. By that logic, Objection is just more speech, and more speech is supposed to be the cure for bad speech.
The Case Against
Then there’s the other side. And it’s a big one.
The $2,000 filing fee is described as accessible. It’s not. For most individuals who feel wronged by a news story, $2,000 is still a significant barrier. But for a corporation, a political campaign, or a billionaire with a grievance, it’s pocket change — less than a decent dinner in Manhattan. As media lawyer Chris Mattei put it: “It seems like a high-tech protection racket for the rich and powerful.”
University of Minnesota media law professor Jane Kirtley sees it as part of a broader pattern of eroding public trust: “If the underlying theme is, ‘Here’s yet another example of how the news media are lying to you,’ that’s one more chink in the armor to help destroy public confidence in independent journalism.”
The system’s evidence hierarchy is where it gets particularly sticky for press freedom. Objection ranks documentary evidence and on-the-record statements at the top. Anonymous sources rank near the bottom. D’Souza has suggested that a “scientific method” approach to journalism would eliminate anonymity altogether — if a source can’t be named, their information shouldn’t count.
This is, to put it gently, a catastrophic misunderstanding of how investigative journalism actually works. Watergate relied on anonymous sources. The Abu Ghraib torture revelations relied on anonymous sources. The #MeToo movement relied on anonymous sources who later went on the record once they felt safe enough. These sources are anonymous not because they’re unreliable, but because they’re vulnerable. An AI tribunal that systematically devalues their testimony isn’t creating accountability — it’s creating a roadmap for powerful people to discredit uncomfortable reporting.
Then there’s the Fire Blanket feature. Posting “under investigation” warnings on X before a verdict is reached is, effectively, a pre-trial punishment. In the attention economy, the accusation arrives before the defence. The narrative gets managed in real time. By the time an “INDETERMINABLE” verdict lands, the damage is already done.
The AI Problem
Setting aside the power dynamics for a moment, there’s a more fundamental question: can LLMs actually adjudicate truth?
The honest answer is: not reliably. These are the same models that have been extensively documented fabricating citations, misreading evidence, and issuing confident-sounding verdicts about things they fundamentally don’t understand. A New York court recently sanctioned a law firm for filing an AI-hallucinated legal brief. Sullivan & Cromwell — one of the most prestigious firms in the world — was caught submitting fabricated case citations generated by AI just last week.
D’Souza has cited a University of Chicago study arguing that AI applies the law with “perfect accuracy.” That study is worth looking at more carefully, because he’s leaning heavily on a finding that doesn’t say what he wants it to say.
The study found that GPT models apply legal rules more consistently than humans — meaning if you give them the same facts ten times, you get the same answer more often. D’Souza translates this to “perfect accuracy.” But consistency and accuracy are different things. A broken clock is consistently wrong twice a day. A model that consistently misapplies a legal precedent is just confident nonsense at scale. The difference matters enormously when the output is a permanent public record attached to a journalist’s name.
There’s also a deeper issue the study doesn’t address: LLMs have no understanding of consequence. A human judge knows that a wrong ruling damages a real person’s life. An LLM doesn’t know what a person is. It can’t weigh the gravity of calling a journalist a liar against the gravity of letting an error stand. It just patterns the next token.
The AI tribunal also only evaluates the evidence submitted to it. It doesn’t know what it doesn’t know. It can’t assess editorial judgment about source protection. It can’t weigh the public interest in publishing against the private harm of keeping something secret. It’s a system designed to evaluate individual factual claims in isolation — which is precisely not how journalism works.
What About the Status Quo?
The strongest argument in Objection’s favour might be the current state of affairs. Traditional media accountability is a mess:
- Defamation lawsuits are expensive, slow, and — in the US — heavily favour the publisher thanks to First Amendment protections and the “actual malice” standard from New York Times v. Sullivan.
- Corrections and retractions are often buried, late, and rarely achieve the same reach as the original error.
- Press councils and ombudsmen have largely vanished from American media.
- Social media pile-ons are the closest thing to real-time accountability, and they’re chaotic, unregulated, and frequently abusive.
Objection is right that this system isn’t working well. But proposing a solution that’s funded by the same billionaire who weaponised the courts against a media outlet, that’s built by the lawyer who executed that strategy, and that systematically devalues the most vulnerable sources in journalism — that’s not fixing the system. That’s building a new weapon for the people who already have the most power in the old one.
If Objection gains traction, it also creates an obvious arms race. A $2,000 filing is trivial for a well-funded adversary but devastating for a newsroom that has to spend staff time responding to dozens of them. It’s not hard to imagine an “AI Defence Council” — a counter-tribunal funded by press freedom organisations, designed to help journalists navigate the process. Which is just privatising the court of public opinion and making everyone pay the AI companies to argue at each other. The lawyers would love it. Everyone else? Probably not.
🔍 The Bottom Line
Objection AI raises a real problem — media accountability is broken — and proposes a solution that makes several specific problems worse. The $2,000 fee creates a two-tier system where the wealthy can challenge stories at will while the people most harmed by bad reporting often can’t afford to. The Honor Index reduces journalistic careers to a numerical score with no room for editorial judgment. The evidence hierarchy systematically disadvantages the most important category of sources in investigative journalism. And the Fire Blanket feature punishes first and asks questions later.
The status quo isn’t great. But replacing a slow, expensive, imperfect system with a fast, cheap, imperfect system run by the people who already have the most power isn’t progress. It’s just a different kind of broken.
The best thing that could come from Objection isn’t the platform itself — it’s the conversation it forces. If it pushes media organisations to be more transparent about corrections, more rigorous about sourcing, and more accountable when they get things wrong, that’s genuinely valuable. But those improvements can happen without handing a billionaire-funded AI a permanent scorecard over every journalist in the world.
For New Zealand, the implications are worth watching. Our media landscape is smaller and more concentrated than America’s. A tool like Objection, deployed by a well-resourced corporation or individual, could exert outsized pressure on local outlets that don’t have the legal or financial resources to fight back. The NZ Media Council exists, but it’s voluntary and its findings are advisory. If Objection-style systems catch on globally, we may need to think about what genuine media accountability looks like — before someone with $2,000 and a grudge decides to think about it for us.
Sources:
