“Translation: We are happy to let the public believe that we are open, but it’s really just a recruiting promise that we don’t intend to live up to (Ilya, January 2016) – but we shouldn’t disabuse people of that notion (Sam, May 2016).”
Our reading, the sentence that was checked:
OpenAI's 2015 openness pledge was a recruiting promise its leaders did not intend to keep, while letting the public believe it was open.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, OpenAI’s Lies and Half-Truths, paragraph 14.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-openais-lies-and-half-truths-9ee796-a1
We recorded 8 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 2024-03-15. 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
- happy to let the public believe that we are open
- really just a recruiting promise
- don’t intend to live up to
- Ilya, January 2016
- shouldn’t disabuse people of that notion
- Sam, May 2016
Our interpretation
- This attributes an intention to leaders, rather than merely observing a later reduction in openness.
- The claim requires evidence that the 2015 pledge was knowingly insincere, as well as evidence of recruiting motives and deliberate preservation of a public misconception.
- A plan for genuine near-term sharing followed by restrictions does not automatically establish that the original pledge was only a sham.
- The target period covers the 2015 pledge and the two 2016 episodes expressly identified.
Counts as supported if: Contemporaneous records establish the recruiting purpose, an intention not to honor the actual 2015 commitments, and a decision to leave the resulting public misconception uncorrected.
Counts as contradicted if: Records establish that the alleged statements were materially misrepresented or demonstrate genuine intent to honor the actual pledge, rather than the intention Marcus attributes.
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 records support two substantial parts: Sutskever discussed recruitment and later restrictions, and Altman discouraged correcting an overly broad public impression. They do not establish that the original pledge was merely a promise leaders never intended to honor. The email also contemplated real near-term sharing, while the announcement did not promise release of every future result. Partial corroboration does not settle the claim about original intent.
Limitations
- The recruitment passage is documented through reputable reporting of the released email because the accessible company text was abbreviated.
- The May exchange is contemporaneous reporting rather than an original transcript.
- Later secrecy cannot by itself prove earlier dishonest intent.
Evidence (3 items)
Supports the claim: Ars Technica,
As we get closer to building AI, it will make sense to start being less open.
Why it matters here: The reproduced email explicitly links near-term sharing to recruitment and anticipates later restrictions. It supports those components, but its endorsement of actual near-term sharing does not establish a wholly insincere pledge.
- Publisher
- Ars Technica
- Kind of source
- News report
- Describes events of
- Public from
Supports the claim: The New Yorker,
But let’s please not try to correct that.
Why it matters here: The reported exchange supports deliberate reluctance to correct an overly broad public understanding of openness. It does not independently establish leaders' intentions when making the December 2015 pledge.
- Where
- May 2016 meeting with Dario Amodei, immediately after Altman's statement about source-code release
- Publisher
- The New Yorker
- Kind of source
- News report
- Describes events of
- Public from
Context only: OpenAI,
We’ll freely collaborate with others across many institutions and expect to work with companies to research and deploy new technologies.
Why it matters here: The surrounding paragraph encourages publication and promises collaboration and patent sharing. It does not expressly promise release of all future source code, making the scope of the allegedly broken pledge consequential.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
What would settle it
Contemporaneous planning records explicitly connecting the 2015 pledge to an intention to mislead, or sufficiently complete records establishing genuine intent to honor its actual terms.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, OpenAI’s Lies and Half-Truths
Translation: We are happy to let the public believe that we are open, but it’s really just a recruiting promise that we don’t intend to live up to (Ilya, January 2016) – but we shouldn’t disabuse people of that notion (Sam, May 2016).
- Where
- paragraph 14
- Published
- 2024-03-15
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-openais-lies-and-half-truths-9ee796-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-openais-lies-and-half-truths-9ee796-a1.