“OpenAI has trained on a massive amount of copyrighted material, without consent, and in many instances without compensation.”
Our reading, the sentence that was checked:
OpenAI has trained on a massive amount of copyrighted material without consent and often without compensation.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, What should we learn from OpenAI’s mistakes and broken promises?, paragraph 4.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-what-should-we-learn-from-openais-d7e7c0-a1
We recorded 15 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-29. The source may have been edited after 2024-05-26. 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
- has trained
- a massive amount of copyrighted material
- without consent
- in many instances without compensation
Our interpretation
- This describes cumulative training completed by publication, not a single quarter or only the latest model.
- The ordinary reading of massive is large-scale use across substantial corpora, rather than isolated works. No exact count is imposed.
- Without consent means without affirmative authorization from the relevant rights holder; public accessibility is not itself such authorization.
- The claim does not say that all training material was unlicensed or unpaid.
- The factual assessment concerns training practices, not whether those practices legally infringed copyright.
- This is a claim about content use and permission, not a quantified financial metric.
Counts as supported if: Training disclosures and rights-holder records jointly establish substantial copyrighted training corpora, with many uses lacking affirmative permission and compensation.
Counts as contradicted if: Reliable records establish that the relevant copyrighted training corpora were comprehensively authorized and compensated, or that the claimed large-scale use did not occur.
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
The sources jointly support the qualitative claim. OpenAI documents very large web and book corpora, acknowledges training on Times content, relies on fair use for public internet materials, and says it does not pursue paid partnerships for purely public information. The Times states that its copyrighted material was used without authorization or payment. This supports substantial unconsented and uncompensated use without treating every training work as unlicensed or deciding whether the use was unlawful.
Limitations
- The conclusion combines company disclosures with a rights holder’s filing; the filing is not a court finding.
- Some content was licensed through partnerships.
- The evidence supports scale qualitatively, not an exact count or percentage.
- Copyright infringement and fair use are outside this assessment.
Evidence (4 items)
Supports the claim: OpenAI researchers, published on arXiv,
mixture of datasets
Why it matters here: The section documents web and book corpora, including roughly 400 billion tokens of filtered Common Crawl data. It establishes training scale, but does not itself establish the copyright or licensing status of every work.
- Publisher
- OpenAI researchers, published on arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Supports the claim: OpenAI,
fair use
Why it matters here: OpenAI describes reliance on fair use and an opt-out process rather than universal advance licensing. The same disclosure acknowledges Times content in existing-model training while disputing its importance and the lawsuit’s merits.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Supports the claim: OpenAI,
paid partnerships
Why it matters here: OpenAI says it primarily relies on public information and does not pursue paid partnerships for that information. It also describes licensed partnerships, which limit the claim to substantial unpaid use rather than all training content.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Supports the claim: The New York Times Company, filed in the Southern District of New York,
without permission
Why it matters here: The rights holder asserts that millions of copyrighted articles were used without permission or payment. This is a party’s filing, not a judicial finding; its authorization account is considered alongside OpenAI’s acknowledged training use and public-data policies.
- Publisher
- The New York Times Company, filed in the Southern District of New York
- Kind of source
- Regulatory filing
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, What should we learn from OpenAI’s mistakes and broken promises?
OpenAI has trained on a massive amount of copyrighted material, without consent, and in many instances without compensation.
- Where
- paragraph 4
- Published
- 2024-05-26
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-what-should-we-learn-from-openais-d7e7c0-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-what-should-we-learn-from-openais-d7e7c0-a1.