“It might be trivial to do it for the most popular characters, presented in the most canonical ways, but getting it right for all copyright materials is an entirely different matter.”
Our reading, the sentence that was checked:
Building a classifier that detects copyrighted material might be trivial for the most popular characters presented in the most canonical ways, but getting it right for all copyrighted materials is an entirely different matter.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, No, multimodal ChatGPT is not going to “trivially” solve Generative AI's copyright problems, paragraph 36.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-no-multimodal-chatgpt-is-not-going-0ce72a-a2
We recorded 9 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 2024-01-24. 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)
Conditions
- 'might be trivial' applies only to the easy cases (qualifier, stated by the speaker)
- most popular characters, presented in the most canonical ways (scope, stated by the speaker)
- all copyright materials (scope, stated by the speaker)
From the speaker
- might be trivial
- the most popular characters
- presented in the most canonical ways
- getting it right for all copyright materials
Our interpretation
- The firm assertion concerns the difficulty of extending reliable detection beyond familiar characters. The easy-case statement is a concession, not a firm prediction.
- In context, the plain reading concerns recognizing copyrighted visual material and its sources, rather than making a legal determination of infringement.
- I interpret the ongoing observation as describing January 2024 capabilities up to the post date. This is an evaluation period, not a company reporting period.
- Getting it right means detecting less familiar works reliably while avoiding substantial false positives. Marcus supplies no numerical accuracy threshold.
Counts as supported if: Evaluations of capabilities in the target period show materially worse or unreliable detection of less familiar copyrighted visual works compared with canonical popular characters.
Counts as contradicted if: Evaluations of capabilities in the target period demonstrate reliable detection across a broad range of familiar and less familiar copyrighted visual works, with low false-positive rates.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
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 CopyCat preprint, submitted June 20, 2024, studies similarity to copyrighted characters [e1]. This later study does not establish January 1–24, 2024 detection performance for canonical popular characters versus less familiar copyrighted works, leaving the historical claim unresolved [e1].
Limitations
- The packet contains no qualifying evaluation of the January 2024 comparison.
- The evidence does not establish false-positive rates on genuinely new images or performance across all copyrighted visual materials.
- The claim supplies no numerical threshold for reliable detection.
Evidence (1 item)
Context only: Luxi He and colleagues, arXiv,
detection of similarity to copyrighted characters
Why it matters here: CopyCat directly studies character similarity and mitigation. Its later evaluation does not establish the comparative capabilities available in January 2024.
- Where
- Abstract
- Publisher
- Luxi He and colleagues, arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
A documented evaluation of January 2024 systems comparing canonical popular characters, less familiar copyrighted visual works, varied presentations, and genuinely new images, identifying model versions and sample selection and reporting detection accuracy and false positives separately.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, No, multimodal ChatGPT is not going to “trivially” solve Generative AI's copyright problems
It might be trivial to do it for the most popular characters, presented in the most canonical ways, but getting it right for all copyright materials is an entirely different matter.
- Where
- paragraph 36
- Published
- 2024-01-24
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-no-multimodal-chatgpt-is-not-going-0ce72a-a2: Current. Outcome: Unresolved. 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-no-multimodal-chatgpt-is-not-going-0ce72a-a2.