"AI Slop" is a 70-page working paper by Jessica M. Silbey and Woodrow Hartzog, both of Boston University School of Law, written December 5, 2025, posted to SSRN on December 8, 2025 and last revised September 6, 2026. It proposes a definition of the term "AI slop" and argues that the concept is useful to policymakers.
This page is written from the SSRN abstract only. The article body could not be retrieved, so the summary below is confined to what the authors state in the abstract; the paper's treatment of its three cultural domains, its policy conclusions and its 47 references are not held. See Provenance.
The proposed definition
The definition the authors offer is expressly provisional: "any output of a probabilistic automated system produced with little exertion that asymmetrically burdens recipients and tends to degrade cultural domains."
Three features are load-bearing in that formulation, and the authors present them as a spectrum rather than a binary test: "Works can be more or less 'sloppy,' depending on the relative presence of its three constituent parts."
| Constituent part | What it picks out |
|---|---|
| Negligible exertion | The output was produced with little effort by the party putting it forward |
| Asymmetrical imposition | The burden of dealing with the output falls on recipients rather than the producer |
| Domain degradation | The output tends to degrade the cultural domain into which it is introduced |
The definition is framed by a question the authors say is unresolved: "What distinguishes slop from other machine-aided human authored work?" Both the provisional status of the answer and the spectrum framing are the authors' own qualifications.
The argument as stated
The abstract describes the term as already carrying "inertia and heft," names its typical referents — AI-generated music and video flooding platforms, low-quality and low-reliability submissions to peer-reviewed journals and legal tribunals, employee reports and internal communications — and states that "the phenomenon has significant consequences. Most of them are not good."
The article's stated method is to interrogate the concept "to identify its essential features and document how it operates in three cultural domains where it is commonly deployed—creativity, research, and work." The three domains are analysed rather than measured; this is a legal-academic argument, not an empirical study.
The conclusion is a claim about the usefulness of a frame rather than about the world: the authors conclude "that AI slop is a useful and important frame in AI discourse and policymaking because it provides a shorthand to distinguish harmful, inferior AI output from meaningful human authorship and because it highlights the pathologies of generative AI for social institutions." See AI Content Saturation (\"AI Slop\"), Media, Journalism & Entertainment — AI Deployment.
The distinction the authors draw between slop and "meaningful human authorship" connects the argument to authorship questions litigated elsewhere — the standard for human authorship in copyright, and the disclosure obligations courts have imposed after AI-fabricated citations. See AI Copyright, Sycophancy and Hallucination.
Provenance and what is missing
The SSRN abstract page was captured complete on September 7, 2026 (HTTP 200). Page metadata carries citation_title "AI Slop", both authors with Boston University School of Law affiliations, citation_publication_date 2025/12/05, citation_online_date 2025/12/08, "70 Pages", "Last revised: 6 Sep 2026", and an empty citation_doi, which is expected for an SSRN working paper.
The article body was not retrievable: SSRN's delivery URL redirects an unauthenticated request back to the abstract page, so no page of the article text beyond the abstract is held. That limits this page to the abstract's own claims, and is why its confidence is low — the constraint is depth of capture, not doubt about authenticity.
Three dates apply and none of them is a September 2026 date of authorship: written December 5, 2025; posted December 8, 2025; revised September 6, 2026. The written date is used as the date of record, with the revision noted, since the argument rather than the upload is what the page describes.
"Slop" is the authors' term, used as theirs. Neither "sloppiness spectrum" nor any similar construction is a term the authors coin, and the concept should not be generalised beyond how they state it.
Relationships
- supports: AI Content Saturation (\"AI Slop\") — supplies a legal-academic definition for a phenomenon otherwise described informally
- related: AI Copyright
- related: AI Content Licensing
- related: Media, Journalism & Entertainment — AI Deployment
- related: Sycophancy and Hallucination