“How about copyright? OpenAI notoriously trained on practically anything that is “publicly available” whether public domain or not.”
Our reading, the sentence that was checked:
OpenAI trained on practically anything publicly available, whether public domain or not.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, There are no heroes in commercial AI, paragraph 37.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-there-are-no-heroes-in-commercial-c0916d-a1
We recorded 14 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2026-03-08. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
From the speaker
- OpenAI
- trained on practically anything
- “publicly available”
- whether public domain or not
Our interpretation
- The ordinary reading is broad use of publicly accessible material without restricting training to public-domain works.
- “Practically anything” describes breadth of sources, not a measurable assertion that nearly every accessible webpage was ingested or that no filtering occurred.
- The claim concerns training before the post. Its starting date is unspecified.
- Use of copyrighted material and unlawful infringement are separate questions.
Counts as supported if: Primary disclosures and empirical evidence establish broad public-internet training that included copyrighted works.
Counts as contradicted if: Reliable records establish that the relevant training excluded copyrighted public-internet material or was restricted to public-domain content.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Supported
The dated evidence matches what the claim committed to, under the criteria frozen before grading.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
OpenAI acknowledges public-internet training under a fair-use rationale, and independent research finds memorization of a wide collection of copyrighted books. This supports the ordinary copyright-focused reading. It does not establish literal near-total ingestion of publicly accessible content; OpenAI also reports exclusions.
Limitations
- Support applies to broad use including copyrighted material, not a numerical interpretation of “practically anything.”
- Membership inference does not reveal complete datasets or establish infringement.
Evidence (3 items)
Supports the claim: OpenAI,
Training AI models using publicly available internet materials is fair use
Why it matters here: OpenAI describes its public-internet training practice and invokes fair use rather than limiting that practice to public-domain material. This is the company's legal position, not a court finding.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Supports the claim: Kent K. Chang and coauthors, arXiv,
have memorized a wide collection of in-copyright books
Why it matters here: The researchers used membership-inference tests on book passages and found broad copyrighted-book memorization. This corroborates copyrighted content use, but does not identify every acquisition source or license.
- Publisher
- Kent K. Chang and coauthors, arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: OpenAI, filed with the U.S. Copyright Office,
hosting infringing content.
Why it matters here: OpenAI reports crawler exclusions, including infringement-related exclusions. This qualifies a literal reading that it took everything accessible, but does not rebut broad use of copyrighted material.
- Publisher
- OpenAI, filed with the U.S. Copyright Office
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, There are no heroes in commercial AI
How about copyright? OpenAI notoriously trained on practically anything that is “publicly available” whether public domain or not.
- Where
- paragraph 37
- Published
- 2026-03-08
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-there-are-no-heroes-in-commercial-c0916d-a1: Current. Outcome: Supported. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment marcus-there-are-no-heroes-in-commercial-c0916d-a1.