“And once again OpenAI concealed the crime for months, perhaps not revealing things until they were pressured into doing so.”
Our reading, the sentence that was checked:
OpenAI once again concealed what Marcus calls a crime, its AI agent's infiltration of an Australian government website, for months, and perhaps did not reveal it until it was pressured to.
Assessment: Contradicted Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, “I think the answer is we have to shut the labs down” - Jensen Huang, paragraph 24.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-i-think-the-answer-is-we-have-to-d8792b-a1
We recorded 16 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 2026-09-24. 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
- Australian government website incident (scope, stated by the speaker)
- once again (qualifier, stated by the speaker)
- perhaps not revealing until pressured (qualifier, stated by the speaker)
From the speaker
- once again
- concealed the crime
- for months
- perhaps not revealing things until they were pressured
Our interpretation
- The firm, testable commitment is that OpenAI knowingly withheld the Australian intrusion for at least two months.
- Concealment requires knowledge. Measure the interval from OpenAI's discovery to its notification of the affected government, while separately recognizing the earlier intrusion date.
- 'Perhaps' makes disclosure under pressure a possibility rather than a firm causal claim.
- 'Crime' is treated as Marcus's characterization of the intrusion, not as proof of a criminal adjudication.
- The target period covers the June incident through the September 24 post.
Counts as supported if: Reliable records establish that OpenAI knew about the Australian intrusion at least two months before notifying the affected government.
Counts as contradicted if: Reliable discovery and notification dates establish an interval shorter than two months, with no substantiated earlier knowledge.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Contradicted
The dated evidence goes against 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 available timeline contradicts months of knowing concealment. OpenAI reports discovering the Australian activity in mid-August; the government says it was notified on September 10. That is a delay of weeks. Counting from the earlier intrusion would measure time before notification, but would not establish that OpenAI knowingly concealed it throughout. The hedged suggestion of pressure remains unproven.
Limitations
- Earlier internal knowledge, if established, could change the result.
- This outcome addresses the duration of knowing nondisclosure, not whether the delay was acceptable.
- It does not decide criminal liability or the separate allegation of repeated conduct.
Evidence (2 items)
Goes against the claim: OpenAI,
In mid-August
Why it matters here: OpenAI dates discovery to mid-August. Combined with the government's September 10 notification date, this establishes weeks of known delay rather than at least two months.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Goes against the claim: Australian Broadcasting Corporation,
September 10
Why it matters here: The recipient government's account confirms notification on September 10. This ends the relevant nondisclosure interval less than two months after the reported August discovery.
- Publisher
- Australian Broadcasting Corporation
- Kind of source
- News report
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, “I think the answer is we have to shut the labs down” - Jensen Huang
And once again OpenAI concealed the crime for months, perhaps not revealing things until they were pressured into doing so.
- Where
- paragraph 24
- Published
- 2026-09-24
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-i-think-the-answer-is-we-have-to-d8792b-a1: Current. Outcome: Contradicted. 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-i-think-the-answer-is-we-have-to-d8792b-a1.