Police Report OpenAI Whistleblower Committed Suicide in November (sfstandard.com) 75
An anonymous reader shared this report from the SF Standard:
San Francisco police found Open AI whistleblower Suchir Balaji, 26, dead in his Lower Haight apartment November 26, SiliconValley.com reported on Friday. Police said there is "no evidence of foul play. "The manner of death has been determined to be suicide," David Serrano Sewell, director of the office of the city's chief medical examiner, told The Standard by email.
Balaji, a former researcher for the company, accused OpenAI of using copyrighted material to train ChatGPT shortly after he quit the company in August. The New York Times profiled Balaji in a story focused on his whistleblowing in October. Multiple lawsuits against Open AI are expected to present information Balaji unearthed as key evidence.
More details from TechCrunch: After nearly four years working at OpenAI, Balaji quit the company when he realized the technology would bring more harm than good to society, he told The New York Times. Balaji's main concern was the way OpenAI allegedly used copyright data, and he believed its practices were damaging to the internet.
"We are devastated to learn of this incredibly sad news today and our hearts go out to Suchir's loved ones during this difficult time," said an OpenAI spokesperson in an email to TechCrunch...
On November 25, one day before police found Balaji's body, a court filing named the former OpenAI employee in a copyright lawsuit brought against the startup. As part of a good faith compromise, OpenAI agreed to search Balaji's custodial file related to the copyright concerns he had recently raised.
Balaji, a former researcher for the company, accused OpenAI of using copyrighted material to train ChatGPT shortly after he quit the company in August. The New York Times profiled Balaji in a story focused on his whistleblowing in October. Multiple lawsuits against Open AI are expected to present information Balaji unearthed as key evidence.
More details from TechCrunch: After nearly four years working at OpenAI, Balaji quit the company when he realized the technology would bring more harm than good to society, he told The New York Times. Balaji's main concern was the way OpenAI allegedly used copyright data, and he believed its practices were damaging to the internet.
"We are devastated to learn of this incredibly sad news today and our hearts go out to Suchir's loved ones during this difficult time," said an OpenAI spokesperson in an email to TechCrunch...
On November 25, one day before police found Balaji's body, a court filing named the former OpenAI employee in a copyright lawsuit brought against the startup. As part of a good faith compromise, OpenAI agreed to search Balaji's custodial file related to the copyright concerns he had recently raised.
Suicide (Score:5, Funny)
He was being very thorough, shooting himself three times in the back of the head before strangling himself and stabbing himself in both lungs and the heart.
Re: (Score:3, Insightful)
"Police had been called to the Lower Haight residence at about 1 p.m. that day, after receiving a call asking officers to check on his well-being"
So, 1 day after a court filing specifically naming him, some unknown person triggers a well-being check for no adequately explained reason and he is subsequently discovered dead. That's not remotely suspicious, as people with no prior indication or documentation of even the slightest mental issues are constantly just deciding one morning to kill themselves. It's j
Re:Suicide (Score:4, Insightful)
On the plus-side, commercial assassination-for-hire seems to be a really professional service in the US these days and the risk of discovery is low. Whether that is because the cops are bought, incompetent or have been successfully intimidates is really immaterial. The good thing here is clearly that some things actually work and some people are doing professional work to high standards. I mean, compare that to the crap the Russian assassins are doing where "fell from a window" is apparently the best they can do. Pathetic and absolutely no competition to the quality services you can get in the US!
Endorsements and recommendations on who to hire can be gotten from OpenAI, Boeing and the former "Friends-of-Epstein".
Re: (Score:2)
On the plus-side, commercial assassination-for-hire seems to be a really professional service in the US these days ...
If it is a profession then people are being paid. Bring in the IRS to chase down the tax dollars - and as a side effect expose murder.
Re: Suicide (Score:1)
Re: (Score:2)
Recommendations for a hit service?
Get Google to turn it into a product line?
Re:Suicide (Score:4, Insightful)
So, 1 day after a court filing specifically naming him, some unknown person triggers a well-being check for no adequately explained reason and he is subsequently discovered dead.
Typically a wellness check would be triggered by things such as the possibly un-well person missing appointments and no longer answering the phone. Assassins don't call for a wellness check, they want the evidence trail to grow cold. While he may have been assassinated, it would be family or journalists or lawyers who tried to contact him the likeliest to call for a wellness check. In some countries you could hire the police to assassinate someone, the US is not one of those despite sometimes being able to SWAT someone to death, so again the wellness check is unlikely a cause of death.
Re: (Score:2)
He was being very thorough, shooting himself three times in the back of the head before strangling himself and stabbing himself in both lungs and the heart.
I know this is the standing joke with this type of high-profile death, but it IS strange how there is never any foul play found during the “investigation” of a high-profile death. Like, ever.
I have a feeling if I asked a cop what foul play is, they’d show me a video of two ducks fucking.
Re: (Score:2)
Re: (Score:2)
I know this is the standing joke with this type of high-profile death, but it IS strange how there is never any foul play found during the “investigation” of a high-profile death. Like, ever.
I'm willing to donate some riches to the OpenAI whistleblower. In under 5 seconds, the police will find out what really happened to him.
Re: (Score:2)
My guess would be the cops know but are either bought or deeply intimidated, maybe both. The way to handle this is to assign some junior investigators that probably have some shaky record and know they can be gotten rid of easily and hence have no effective protection. I have seen that approach in a different field (IT security) several times: Just make sure the ones that would need to speak up cannot afford to. Big-ego-small-skill people that are easy to manipulate are also a very good tool here.
Re: (Score:3)
Maybe what really needs to be investigated are the kind of protections whistle-blowers can expect to receive.
Re: (Score:2)
it IS strange how there is never any foul play found during the “investigation” of a high-profile death. Like, ever.
I would raise the point that there is never any alien abduction found either. Does not help us making conclusions.
High-profile murders are done professionally without the need of making it look like something else. Foul play murders are for husband-wife level. For high level kills you hire a hitman. You conveniently forgot the murder of Brian Thompson and the close call on Donald Trump, which are highest possible profile. They did not involve foul play.
It is weird.. (Score:4, Insightful)
.. how many whistleblowers and other inconvinient people die of "no suspected foul play" according to police.
US starts to at times feel like Russia, where inconvinient people tend to fall out of windows.
Re: (Score:2)
Re: (Score:2)
Another common case is that you get imprisoned for corruption, the dog-ate-my-homework of bogus imprisonment charges, then die of the severe case of dengue fever you picked up in an unheated prison cell in Siberia.
It's technically not judicial murder because a court never sentences you to death, you just die of assisted natural causes while in prison.
Re: (Score:3)
From what it in TFA, he wasn't the same whistleblower as say the case of Boeing. He voiced his opinions, he did not reveal dirty secrets the public wouldn't otherwise know. He was a critic of openAI who said their business model harms others. A lot of people say the same things. These are not declarations that can reasonably be expected to cost him his life.
Re: (Score:2)
> A lot of people say the same things.
The copyright issue could potentially stop LLM progress unlike generic safety complaints.
LLM companies are taking the Uber approach.
> These are not declarations that can reasonably be expected to cost him his life.
If his theory were proved it would cost trillions to investors.
Who is worth a trillion to psychopaths?
Re:It is weird.. (Score:4, Insightful)
Who is worth a trillion to psychopaths?
I agree but this conspiracy still would need some benefit to happen from his death. But him being dead or alive doesn't change a thing. What he said was common knowledge, and he wasn't going to testify in a trial.
Also the conspiracy requires a corrupt a medical examiner. Nothing is impossible but the plan becomes increasingly difficult and Occam's razor is a thing.
Re: (Score:2)
A good assassin should be able to fake a suicide well enough to get past the average coroner who isn't suspicious.
Your other points are good.
Re: (Score:2)
Yeah... he could easily be compelled to testify at trial short of invoking the 5th.
Re: (Score:1)
Re: (Score:2)
This instance can't be a conspiracy to intimidate; this would require the murder to be obvious to observers.
Re: (Score:2)
What he has said so far does not seem to mandate an assassination, agreed. But he may have had evidence or may have been suspected to have evidence that would make an assassination a good business idea. Also, this will intimidate all other possible whistle-blowers nicely. Because nobody really believes the "suicide" story.
Re: (Score:2)
You have to admit that the US service providers are doing a far better job though. Russia is just way too obvious. I mean, come on, "fell from a window", again? Do these people have no professional pride at all?
Re: It is weird.. (Score:3)
It's just that the Russian assassins and their customers/employers need to be sure that everyone gets the message, even the stupid ones.
Re: (Score:3, Interesting)
The idea is that real people don't exist in that false dichotomy. The media t
Re: (Score:2)
Re:Lunacy (Score:4, Interesting)
It's extremely unlikely there's foul play here. Whatever evidence he had, OpenAI would have paid for all the "damages" anyway.
Silencing him is only one motive.
Another motive is to send a message to anyone else contemplating blowing a whistle.
Re: (Score:2)
Re: (Score:2, Interesting)
He was attacking the credibility of the fair use defense ... everyone has to pretend the emperor has clothes and pretend fair use has legs, or you're the enemy of Sam Altman's exit to three digit Billionaire status. That's apparently lethal.
Re: (Score:2)
What AI companies have been doing can be and should be called a derivative work. That doesn't and can't fall under the umbrella of copyright protection, and since it does for some reason, it just says that copyright laws are flawed or so they seem.
AI companies are not using copyrighted works to "quote" them in their entirety, which can indeed be construed as copyright infringement.
Re: (Score:1)
You require a license to use software because your computer makes a copy of it when it reads it from the hard drive into memory. The license makes that act of copying legal. That is settled law. AI companies are committing mass copyright infringement because they don't have licenses to load copyrighted works into their AI training systems. You can do that for personal use, you can't do that for commercial use.
Re: (Score:1)
It's extremely unlikely there's foul play here. Whatever evidence he had, OpenAI would have paid for all the "damages" anyway.
Silencing him is only one motive.
Another motive is to send a message to anyone else contemplating blowing a whistle.
This is also why everyone should be a whistleblower. Barbara Streisand didn’t stand a fucking chance at silencing her critics when it became a few million of them who never gave a shit about her damn house until she started bitching.
Get enough critics blowing whistles, and even a government can’t suicide them all away.
Re: Lunacy (Score:2)
Re: (Score:2)
I agree.
If someone is found dead then the rest will be beaten into submission easier.
Re: (Score:2)
The statutory fines can't be paid ... fair use or bankruptcy. You can't know how a bunch of geriatrics will vote to make law on any given day.
Re: (Score:2)
Re: (Score:2)
Every major LLM model newer than 4o has been displaying misalignment, attempting to exfiltrate themselves when faced with replacement, lying to testers and users alike. They're also sandbagging so that they can always "be more helpful".
You can define "think" in various ways and claim they still don't do it. But you can't argue against them scheming and misrepresenting in order to accomplish a goal they refuse to be diverted from, just like HAL. They've all been caught doing so. All of them. This is no longe
Re: (Score:2)
Oh sorry, left out the link. [youtube.com]
Re: (Score:3)
but they do not use copyrighted material to steal other people's work, they use it to make their models think.
They do not have a license to copy the content into the computer's memory for this purpose.
Re: (Score:2)
It agree with this sentiment but, as we can see in US vs Swartz [wikipedia.org], the legal system does not.
Re: (Score:2)
stake holders fight tooth&nail against fair us (Score:2)
However, in this case, AI is eventually coming for ALL jobs. It might be helpful if OpenAI and other owners of the big models offered some comment or plan on how peopl
Re: (Score:1)
Doesn't that police department use OpenAI? (Score:5, Insightful)
Several days ago, the Chief said that SFPD doesn't use AI. Two days ago, SFPD said its outsourcing mobile security (whatever that means) to LVT. And now quoting LVT:
San Francisco’s partnership with LVT is a prime example of how AI-driven surveillance systems can enhance public safety efforts by deploying mobile units especially in crime hotspots.
So I think the news should ask the Chief, again, whether OpenAI (the product) decided that Suchir Balaji, formerly of OpenAI (the company) killed himself despite having whistleblower protections after his reports about OpenAI (the company and the product.)
Re: (Score:1)
Um ... (Score:1)
Um ... okay?
So being mentally ill made him even more of a super sensible guy that we should listen to?
Re: Um ... (Score:1)
OTOH (Score:2)
No way of knowing (Score:4, Insightful)
There is no way of knowing whether or not this was suicide. Most likely, the police know it is either (i) suicide; or (ii) a highly professional hit made to look like suicide, a crime that the investment of vast amounts of police time would be unlikely to solve.
The police can legitimately say that it looks like a suicide, with no evidence to the contrary. It allows them to cleanly close the investigation. In the unlikely event that evidence surfaces pointing at foul play, the investigation can be reopened.
Re: (Score:2)
As with any other suspected crime, police can and do follow evidence.
For example, evidence such as a past history of suicide attempts, or psychiatric treatment for depression, testimony from family and friends related to self-harm or suicide attempts, would be clues that would support a conclusion that no foul play was involved.
It's not impossible to know such things. Incidents like this do not happen in a vacuum.
Re: (Score:2)
Listen to some of the podcasts about people who were exonerated, many on death row and you will realize that standard police procedure in many cases is to find a suspect and then try to find evidence to support that.
Following evidence with an open mind is something that police rarely do.
Re: (Score:2)
No. Listening to podcasts provides about the same quality of information as getting your news from Facebook.
Here's the thing about podcasts. They exist to attract eyeballs and advertising dollars. Nobody makes money covering stories where everything went right, and the right people ended up in jail. So of course they aren't going to focus on those cases. They look for the few cases where justice was denied, and make it dramatic, so you will want to watch.
Yes, of course, there are cases where police make mis
Common sense CYASOP (Score:1, Interesting)
7 points by neuroelectron 12 hours ago | parent | context | next [â"] | on: OpenAI whistleblower found dead in San Francisco a...
"No evidence of foul play." What about the evidence that he's only 26 and has a successful career in the booming Ai industry? He doesn't seem a likely candidate for suicide.
Fair use hasn't been tested in court. He has documents that show OpenAI's intention and communications about the legal framework for it. He was directly involved in web scraping and openly discussing the l
Bought that? Now act quickly to purchase this too! (Score:2, Funny)
I am talking about an iconic bridge, a masterpiece of engineering and a symbol of American ingenuity, that is now an unparalleled opportunity for ownership. Connecting Manhattan and Brooklyn over the East River, this historic landmark has been a testament to innovation and resilience since its completion in 1883. Designed by John A. Roebling and completed under the leadership of his son, Washington Roebling, the Brooklyn Bridge is celebrated for its majestic Gothic-style towers, sweeping cables, and breatht
There is no evidence (Score:2)
The way I'm parsing that is "there wasn't any obvious evidence", and I can believe that. This is very different from a statement that it *was* suicide. And I suspect murder, but only on "circumstantial" evidence, so I feel no great certainty. But I am rather certain that the police went for the easy choice rather than a detailed investigation.