“OpenAI will no doubt by now have filtered this specific example so that it can no longer be replicated, now that it has appeared in a lawsuit.”
Our reading, the sentence that was checked:
OpenAI has no doubt by now filtered this specific NYT-lawsuit example so that it can no longer be replicated, now that it has appeared in a lawsuit.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Deconstructing Geoffrey Hinton’s weakest argument, paragraph 52 (footnote 2).
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-deconstructing-geoffrey-hintons-weakest-5731dc-a1
We recorded 10 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 2024-02-05. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- Deadline, as the speaker wrote it
- “by now”
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
Conditions
- no doubt (qualifier, stated by the speaker)
- now that it has appeared in a lawsuit (qualifier, stated by the speaker)
From the speaker
- OpenAI will no doubt by now have filtered
- this specific example
- so that it can no longer be replicated
- now that it has appeared in a lawsuit
Our interpretation
- By now asserts a completed change by February 5, 2024. Despite the future-perfect phrasing, this is an observation about the state at publication, not a future forecast.
- This specific example refers to the lawsuit reproduction illustrated in the post, not every New York Times reproduction or every memorization attack.
- Support requires evidence both of a filtering change affecting that example and of its effect on reproducibility. A general anti-regurgitation policy is insufficient.
- The target is the deployed system's state on February 5, 2024. Earlier statements and later tests are context unless they establish that historical state.
Counts as supported if: A dated technical disclosure, deployment record, or adequately controlled contemporaneous test establishes that OpenAI had blocked reproduction of the illustrated example by February 5, 2024.
Counts as contradicted if: A documented successful replication on the relevant deployed system at the target date, or reliable records establishing that the asserted filter was absent or ineffective then.
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
OpenAI had publicly described general anti-regurgitation measures before the post. That does not establish that it filtered the specific illustrated example or made it irreproducible by February 5. The searches and reviewed filing did not supply the necessary change record or contemporaneous replication test. The confident wording is therefore unverified, rather than disproved.
Limitations
- The relevant generation setup is unspecified.
- A filter could have existed without a public announcement.
- Later model behavior cannot by itself establish behavior on the target date.
Evidence (1 item)
Context only: OpenAI,
measures in place
Why it matters here: OpenAI publicly described general measures against memorization and regurgitation. This does not identify Marcus's example, a change prompted by the lawsuit, or the system's reproducibility on February 5.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
What would settle it
Identify the exact example and obtain dated deployment records or contemporaneous tests for the relevant model and interface. A successful replication at the target date would directly challenge the claim; support requires evidence of the asserted blocking change.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Deconstructing Geoffrey Hinton’s weakest argument
OpenAI will no doubt by now have filtered this specific example so that it can no longer be replicated, now that it has appeared in a lawsuit.
- Published
- 2024-02-05
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-deconstructing-geoffrey-hintons-weakest-5731dc-a1: 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-deconstructing-geoffrey-hintons-weakest-5731dc-a1.