Two Accountability Arguments, Zero Contact
The Anthropic federal case is being shaped by a framework that the people most invested in AI accountability have never encountered. Fourteen Catholic moral theologians filed amicus briefs [15] engaging the central liability question — who bears responsibility when an algorithm contributes to harm — through a tradition of moral philosophy that has been working on questions of proportionality and institutional responsibility for centuries. Pope Leo XIV has made AI ethics a stated priority of his papacy, and a Vatican body's assessment has been described as the International Theological Commission's stark warning — language strong enough to suggest the moral judgment has already been delivered. The legal argument is the implementation of that judgment.
The Bluesky conversation happening in parallel is not ignorant of the stakes — it is simply running on a different channel, with a different institutional form. Posts characterizing AI corporations as structurally designed to avoid accountability [4] are making the same structural claim the amicus briefs make. But a structural claim made in a social feed and a structural claim made in a federal amicus brief have different trajectories. Only one of them is in the room where the liability framework gets written.
What the Theological Framework Actually Claims
The conference at Rome's Pontifical University of St. Thomas Aquinas asked whether AI can be ordered toward virtue — a question that sounds, from outside the tradition, like corporate ethics marketing. Inside the tradition, it is the opposite: the Thomistic framework requires that virtue be attributable to agents who make choices, which means that when a system causes harm, the chain of moral responsibility runs back through every human decision that shaped it. The doctrine does not allow corporations to disappear behind technical complexity. It demands that the individuals who designed the system, made the deployment decisions, and profited from the outcomes be morally — and by extension, legally — traceable.
This is why the theologians' amicus briefs support Anthropic in a narrow procedural sense while advancing a principle that limits every AI company's liability shield. The argument is not that Anthropic is virtuous. The argument is that the question of virtue is always attributable to specific humans, not to the corporation or the algorithm as abstractions. That principle, if adopted, is the enforcement mechanism that the Bluesky conversation has been asking for without knowing the name of the doctrine that would provide it.
The Calcification Problem on the Skeptic Side
The coherence of Bluesky's AI critique has a structural cost. When the dominant mode of expression is "ai bad" [14] and "there is no such thing as ethical AI" [10], the conversation forecloses precisely the distinctions that would make accountability arguments legally actionable. The theological framework requires holding two things simultaneously: that AI systems cause harm, and that the harm is attributable to specific human agents rather than to the technology as such. The first half is Bluesky consensus. The second half — which is the part that can be litigated — requires treating the ethics question as still open, which is the position the conversation has officially rejected.
The irony is that the rejection of "ethical AI" as a category serves the corporations the community most wants to hold responsible. If accountability requires showing that specific humans made specific decisions that caused specific harm, then the argument that no AI use case is ethical collapses the distinctions those humans need to be named. A scholar in the conversation acknowledged the more tractable version of the problem — that AI is entering the infrastructure of scientific publishing and requires accountability mechanisms within that infrastructure [5] — but that framing is not the one that sets the tone. The tone is set by the posts that travel, and the posts that travel have already delivered a verdict.
The Debate Happening Without an Audience
There is a debate inside Catholic intellectual circles about whether the theological framework endorses AI or constrains it — a debate between a Magisterium AI founder and a critic over whether conversing with a Catholic-trained chatbot is a genuine encounter with truth or a fundamentally disordered act. That debate is more sophisticated than most of what the Bluesky conversation produces, and it is happening in podcast format, reaching an audience that is not the one most angry about AI's harms.
The communities most invested in AI accountability — artists, labor organizers, critics of corporate capture — are not in the room where the most philosophically developed accountability arguments are being made. "AI content generation ethics is a hell of a grift" [13] is a response to corporate ethics theater, and it is not wrong about that target. But it mistakes the theological accountability argument for the same grift because the two look similar from a distance: both involve the word ethics, both involve institutions with resources, both produce documents. The difference is that one set of documents is being filed in federal court.
The Argument That Reaches Court
The practical consequence is not abstract. If the theological framework — that moral and legal responsibility must be traceable to the individuals behind the algorithms — is adopted by the court, it becomes the enforcement template for AI liability cases that follow. The grassroots demand that "we start going after the individuals behind the algorithms" [4] would have its legal mechanism, written by people who were never part of the conversation that generated the demand.
The communities most invested in that outcome are not positioned to influence how the argument is framed, because they have defined the ethics conversation as captured territory. The scholars advancing the accountability argument in federal court are the least likely voices to be engaged by the communities whose interests they are advancing. The accountability framework that lands will be written by Dominican friars and Catholic philosophers — and the people who needed it most will have spent the decisive period arguing that no such framework was possible.