“The company ripped off a lot of writer’s work (per the $1.5B settlement), and recently walked back its core safety pledge.”
Our reading, the sentence that was checked:
Anthropic ripped off a lot of writers' work, as shown by its $1.5B settlement.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, The whole thing was a scam, paragraph 12.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-the-whole-thing-was-scam-c984d6-a2
We recorded 5 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 2026-02-28. 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
- The company
- ripped off a lot of writer’s work
- per the $1.5B settlement
Our interpretation
- The company means Anthropic.
- The ordinary reading of 'ripped off' is that Anthropic obtained unauthorized copies of many writers' books. It does not require a criminal conviction or establish that all AI training infringes copyright.
- The settlement reference identifies the Bartz litigation and asserts an approximately $1.5 billion settlement agreement, not that this amount had already been paid.
- This is principally a copyright-conduct claim. The settlement amount identifies the associated legal resolution, so metric_scope is null.
- The claim concerns past conduct and its settlement before publication. No beginning date is supplied, so the target period has an unspecified start.
Counts as supported if: Court records establish substantial unauthorized copying of writers' books by Anthropic, and the related litigation produced an approximately $1.5 billion settlement agreement before the post.
Counts as contradicted if: The records establish that the relevant copying was authorized or did not occur on a substantial scale, or that the cited settlement did not exist or concerned materially different conduct.
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 court's October 17, 2025 opinion describes Anthropic downloading millions of books from pirate libraries into its central library and says that acquisition was not fair use on the summary-judgment record [e1]. The settlement notice filed October 1, 2025 confirms an agreed $1.5 billion fund [e2]. These records support substantial unauthorized book acquisition and the associated settlement agreement [e1, e2]. The settlement alone does not establish wrongdoing: liability remained disputed and the claims had not been tried [e2]. The June 23, 2025 order held the training use to be fair use, so the evidence does not support treating that use as unlawful [e3].
Limitations
- The assessment reads 'ripped off' as unauthorized acquisition, not unlawful training.
- The court accepted the authors' version of disputed piracy facts at summary judgment; this was not a final liability verdict [e3].
- The evidence does not establish an exact number of distinct writers.
- The $1.5 billion figure describes an agreed fund, not an amount already distributed to authors [e2].
Evidence (3 items)
Supports the claim: U.S. District Court, Northern District of California,
Anthropic had copied many pirated books.
Why it matters here: The court describes substantial copying and identifies 482,460 settlement works in footnote 2. The underlying LibGen and PiLiMi acquisitions occurred in 2021 and 2022, as detailed in the June opinion cited in e3.
- Publisher
- U.S. District Court, Northern District of California
- Kind of source
- Regulatory filing
- Describes events of
- Public from
Supports the claim: Parties to Bartz v. Anthropic; filed in the U.S. District Court, Northern District of California,
Under the Settlement, Anthropic has agreed to establish a non-reversionary Settlement Fund of one-billion-five-hundred-million dollars ($1,500,000,000).
Why it matters here: The filed notice confirms the agreed $1.5 billion fund. Question 4 also states that the parties still disagree over liability and that the claims had not been tried, so the settlement is not itself proof of wrongdoing.
- Publisher
- Parties to Bartz v. Anthropic; filed in the U.S. District Court, Northern District of California
- Kind of source
- Regulatory filing
- Describes events of
- Public from
Context only: U.S. District Court, Northern District of California,
This order grants summary judgment for Anthropic that the training use was a fair use.
Why it matters here: The court upheld the training use while treating the pirated library acquisitions separately. Its conclusion also explains that disputed facts concerning piracy were viewed in the authors' favor because Anthropic sought summary judgment; this was not a final trial verdict.
- Publisher
- U.S. District Court, Northern District of California
- 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, The whole thing was a scam
The company ripped off a lot of writer’s work (per the $1.5B settlement), and recently walked back its core safety pledge.
- Where
- paragraph 12
- Published
- 2026-02-28
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-the-whole-thing-was-scam-c984d6-a2: 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.