What should we learn from OpenAI’s mistakes and broken promises?
We recorded 15 claims from this post. 7 were checked against evidence. 8 can’t be checked. 3 more are held. Read the post at its source.
Claims from this post (15)
In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.
Assessment: UnresolvedSaid
“But lately, he and they have made a lot of mistakes, and broken a lot of promises.”
What we tested: Lately Sam Altman and OpenAI have made many mistakes and broken many promises.
Assessment: SupportedSaid
“OpenAI called itself open, and traded on the notion of being open, but even as early as May 2016 knew that the name was misleading.”
What we tested: OpenAI traded on being 'open' while knowing since May 2016 that the name was misleading.
What happened: The report describes Amodei telling Altman and Brockman that people expected OpenAI to release its source code. Altman acknowledged that this exceeded its plans and discouraged correcting the impression; this is firsthand reporting, not an internal email. (The New Yorker, )
Assessment: MixedSaid
“OpenAI proceeded to make a Scarlett Johansson-like voice for GPT-4o, even after she specifically told them not to, behaving in an entitled fashion that has angered many—highlighting their overall dismissive attitude towards artist consent.”
What we tested: OpenAI made a Scarlett Johansson-like voice for GPT-4o after she told them not to, showing dismissiveness toward artist consent.
What held: The report supports recognizable resemblance in the public demonstration. It does not establish cloning or deliberate imitation. (NPR, carried by Iowa Public Radio, )
What did not: OpenAI dates recording to June and July 2023 and Johansson’s first approach and refusal to September. This contradicts creation of the original Sky voice after her refusal, although deployment followed it. (OpenAI, )
Assessment: Can’t be checkedSaid
“Millions, perhaps billions, of documents likely have been taken without permission.”
What we tested: OpenAI has likely taken millions, perhaps billions, of documents without permission.
Assessment: SupportedSaid
“OpenAI has trained on a massive amount of copyrighted material, without consent, and in many instances without compensation.”
What we tested: OpenAI has trained on a massive amount of copyrighted material without consent and often without compensation.
What happened: The section documents web and book corpora, including roughly 400 billion tokens of filtered Common Crawl data. It establishes training scale, but does not itself establish the copyright or licensing status of every work. (OpenAI researchers, published on arXiv, )
Assessment: UnresolvedSaid
“OpenAI promised to devote 20% of its efforts to AI safety, but never delivered, according to a recent report by Jeremy Kahn at Fortune.”
What we tested: OpenAI never delivered on its promise to devote 20% of its efforts to AI safety.
Assessment: SupportedSaid
“Altman once (in 2016) promised that outsiders would play an important role in the company’s governance; that key promise has not been kept.”
What we tested: Altman's 2016 promise that outsiders would play an important role in OpenAI governance has not been kept.
What happened: The adjacent line 7a explanation assigns director elections to these members. The filing therefore describes an internally selected board, not an electorate drawn from the public. (OpenAI Inc., copy hosted by Gwern, )
Assessment: Can’t be checkedSaid
“The list above is not complete, but it shows a significant gap between the rosy rhetoric and actual reality, and in my view that gap has only grown over time, particularly over the last year or so..”
What we tested: The author's non-exhaustive list shows a significant gap between OpenAI's rosy rhetoric and reality, and in his view that gap has only grown over time, particularly over the last year or so.
Assessment: Can’t be checkedSaid
“At this point, I think it is correct for the public to take everything OpenAI says with a grain of salt.”
What we tested: The public should take everything OpenAI says with a grain of salt.
Assessment: Can’t be checkedSaid
“But OpenAI still have massive power, and with it a chance to screw up, and possibly put humanity at risk.”
What we tested: OpenAI has massive power and could possibly put humanity at risk.
Assessment: Can’t be checkedSaid
“Our government has been repeatedly complicit, treating Altman repeatedly as royalty.”
What we tested: The US government has repeatedly been complicit, repeatedly treating Altman as royalty.
Assessment: Can’t be checkedSaid
“If the first lesson is specific - that OpenAI should not be fully trusted – the second lesson is general.”
What we tested: OpenAI should not be fully trusted.
Assessment: UnresolvedSaid
“Regulatory capture is the default, as Karla Ortiz recently noted,::”
What we tested: Regulatory capture is the default outcome in AI regulation.
Assessment: Can’t be checkedSaid
“Without independent scientists in the loop, with a real voice, we are lost.”
What we tested: Without independent scientists in the loop with a real voice, we are lost.
Assessment: Can’t be checkedSaid
“As I argue, in my upcoming book Taming Silicon Valley, our only real hope is if citizens speak up.”
What we tested: The only real hope for governing AI is citizens speaking up.
Held from this post (3)
These records did not pass every check, so nothing about them is published. Held is not a finding.
- The criteria are ambiguous. Our reading, not assessed: Sam Altman appears to have misled people, including US Senators, about his personal holdings in OpenAI: he correctly reported having no direct equity but did not mention his ownership in the OpenAI Startup Fund or his indirect holdings in OpenAI through YCombinator.
- The two reviews read the claim differently
- The record does not meet the current rules. Our reading, not assessed: As recently as November 2023, OpenAI filed for nonprofit tax-exempt status saying its mission was to safely benefit humanity unconstrained by a need to generate financial return, even though it turns over almost half its profits, if it makes any, to Microsoft and seems to have back-burnered its safety work.