OpenAI is SERIOUSLY Concerning Me…

It's been an interesting couple weeks for OpenAI. They've really sort of lost a lot of the trust of the general public who's paying attention to what OpenAI is doing. I mean, who needs to watch drama on TV anymore when the most fascinating drama on the planet is happening over at OpenAI and in the general AI world right now? And I'm going to break it all down for you, starting with some semi bombshell news coming from my good buddy Bilavel Sidhu, who was recently on the Next Wave podcast. By the way, he has a podcast called the Ted AI Podcast. And he had the opportunity to interview Helen Toner. Helen Toner was one of the board members back in November of last year when all of the crazy stuff went down over at OpenAI. If you remember, Sam Altman was fired by the board, and then the board was essentially removed, and then Sam Altman was brought back in. Well, Helen Toner was one of those people that was on the board that. That voted to get rid of Sam. She was also one of those board members that was promptly removed from the board when Sam came back. Well, for the first time since November, when all of that went down, Helen finally went on a podcast to talk about her side of the story. And probably one of the most interesting parts of this conversation was this bit.

Right here to give a sense of sort of the kind of thing that I'm talking about is things like, you know, when ChatGPT came out November 2022, the board was not informed in advance about that. We learned about ChatGPT on Twitter.

So according to Helen Toner, the board at OpenAI didn't find out about the launch of ChatGPT until they read it on Twitter. Now, I find this really interesting because at the time, the board consisted of six members. It had Sam Altman, the CEO of OpenAI, Greg Brockman, the president of OpenAI, Ilya Suskever, one of the lead engineer scientists at OpenAI. And all three of these people definitely knew ChatGPT was coming out. And then you had Helen Toner, Tasha McCauley and Adam D', Angelo, three people that didn't actually work at OpenAI but were on this board. So it wasn't really that the board didn't know ChatGPT was coming. It was more that half the board was unaware that ChatGPT was coming, and the other half never told them about it. The other thing that's interesting, that's pointed out by Sean Ralston here, who does OpenAI API dev support, he says this is ridiculous. Chat GPT was well known, tested and the API was being used by many vendors and third party chatbots. So the GPT3 technology underneath chat GPT was already being used by a lot of companies, including some really big ones like Jasper8a PI here points out it was a research release of a literal GPT3 wrapper of an API that Jasper had been using publicly for like a year. There was no new model, no new counterpart, no new contract signed, no obligation entered into. So the idea here that these guys are trying to point out is that maybe the whole board wasn't aware of the date that ChatGPT was going to be launched, but they were definitely aware of the technology and how it worked and there was already some of this technology being used in the wild. And when ChatGPT first came out it was meant to be a research preview. I don't believe OpenAI expected millions of people to jump on board within weeks of its launch. So I do want to sort of disclaim that just to give a little additional credit to the OpenAI side. But still, it does sound like at least half the board was intentionally left out of the loop about what was going to happen with the company that they have a responsibility to oversee and sort of keep safe. But that's just a tiny piece of the hot water we've been seeing OpenAI get themselves into recently. A few weeks ago, Jan, the head of alignment over at OpenAI, stepped down from his role at OpenAI, saying that I've been disagreeing with OpenAI leadership about the company's core priorities for quite some time, until we finally reached a breaking point. Building smarter than human machines is an inherently dangerous and endeavor. OpenAI is shouldering an enormous responsibility on behalf of all humanity. But over the past years, safety, culture and processes have taken a backseat to shiny products. Now when Jan spoke up like this, it raised some more questions like why has everybody else who left OpenAI recently not been speaking out about the dangers and the fact that maybe OpenAI is approaching things in a less than safe way. Well, it turned out that OpenAI had an agreement that people signed when they left the company where they agreed to a non disparagement, meaning they couldn't talk badly about OpenAI and if they did, they would lose their vested equity in the company. I mean that borderline doesn't even feel legal. However, Sam Altman took to Twitter to say in regards to recent stuff about how OpenAI handles equity, we've never clawed back anyone's vested equity, nor will we do that if People do not sign a separation agreement. Vested equity is vested equity, full stop. He goes on to say, it should never have been something we had in any documents or communication. This is on me. And one of the few times I've been genuinely embarrassed running OpenAI. I did not know this was happening and I should have. And that line right there is important. More and more evidence came out that Sam actually did know that this was part of it and he agreed with it. In fact, if we go back and look at this post on Sam Altman, there's a community post on it saying, according to leaked messages and documents obtained by Vox, senior leadership at OpenAI, including Altman, were very much aware of these equity clawback provisions and signed off on them. Here's the article in question from Vox. Leaked OpenAI document reveal aggressive tactics towards former employees. In the article, this says company documents obtained by Vox with signatures from Altman complicate their claim that the Club act provisions were something they hadn't known about. This Vox article also publicly published all of the agreements that people had to sign so that you can actually review them. But you can clearly see that Sam Altman did sign off on all of it. Now, what about OpenAI safety? Obviously, Helen Toner and the early board members had some concerns and they were booted off the board. And then we had Jan, who was the head of Alignment, and he left and they disbanded the Alignment team. So where does OpenAI stand right now? Now, don't worry about that, because OpenAI formed a new Safety and Security Committee. This new committee is responsible for making recommendations on critical safety and security decisions for all OpenAI projects. Recommendations in 90 days. So, of course, they're probably going to bring in people from outside of OpenAI to make sure that the people inside of OpenAI are doing things in a safe way. Right. Today, the OpenAI board formed a Safety and Security Committee led by directors Brett Taylor, Adam d', Angelo, Nicole Seligman and Sam Altman. This committee will be responsible for making recommendations to the full board on critical safety and security decisions for OpenAI projects and operations. So you're telling me that the CEO of OpenAI, who makes all of the financial decisions on what to launch and is basically the salesman for OpenAI, is also the guy making the decisions on the security and safety of OpenAI. Seems a little conflict y. Now, the board also includes a whole bunch of other people, but all of them are OpenAI employees. And if you're wondering about Jan, the guy who used to run alignment over at OpenAI, well, this week it was announced that he went and joined Anthropic to head up their new Alignment team. Jan took to Twitter to say, I'm excited to join Anthropic AI to continue the super alignment mission. My new team will work on scalable oversight, weak to strong generalization, and automated alignment research. Now, Anthropic is one of those companies that I actually do kind of trust. I kind of sort of trusted OpenAI for a while, but I mean, they're kind of getting more and more on shaky ground with each news article that comes out. Anthropic has been one of the few companies that has consistently felt like alignment and safety has been their number one priority. And that narrative has always kind of been there with Anthropic. So I really do think that Jan's landing in a good spot going to Anthropic. Look, we're just scratching the surface right now of some of the troubles that OpenAI has been battling right now. We also had this little issue where the cto, miramoradi, couldn't really answer the question about how the SORA model was actually trained.

What data was used to train sora? We used publicly available data and licensed Data videos on YouTube. I'm actually not sure about that. Videos from Facebook, Instagram, you know, if they were publicly available, available yet publicly available to use. There might be that data, but I'm not sure. I'm not confident about it. What about Shutterstock? I know you guys have a deal with them. I'm just not going to go into the details of the data that was, that was used, but it was publicly available or licensed data.

Well, as it turned out, according to this article from the New York Times, OpenAI transcribed over a million hours of YouTube videos to train GPT4. Now, this isn't referencing Sora. Mira was being asked about Sora. This New York Times article is saying GPT4 was trained with YouTube videos. But I don't think it's too much of a leap to assume that if they're willing to train GPT4 with YouTube videos, they were probably willing to trade Sora with YouTube videos as well. And then we, we have the Scarlett Johansson issue that's recently popped up. After the GPT4 Omni demo, Scarlett Johansson actually hired a lawyer to deal with OpenAI because she believed that OpenAI used her voice without permission to create the sky voice model for GPT4. O Johansson said she hired legal counsel and said OpenAI reluctantly agreed to take down the sky voice after her council sent Altman two letters. The OpenAI team put up a post over on the OpenAI blog saying how the voices for ChatGPT were chosen. He claims that they hired different voice actors and never intended to make it sound like Scarlett Johansson's voice. You can see here the voice of sky is not Scarlett Johansson's and it was never intended to resemble hers. We cast the voice actor behind Sky's voice before any outreach to Ms. Johansson. Out of respect for Ms. Johansson, we have paused using Sky's voice in our products. And the fact that he says it was never intended to resemble hers and is what I find interesting. Because if it was never intended to resemble hers, why did Sam tweet just the word her 24 hours before the GPT4O event? Obviously, he was referencing the movie her, which was about an AI that was voiced by Scarlett Johansson. And if he never intended it to sound like Scarlett Johansson, why did he reach out and ask if she would voice it in the first place? It's just not adding up. And now, if we go all the way back to November 17, 2023, when OpenAI announces a leadership transition, the reason given for firing Sam Altman was that he was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities. And I think given everything that we've seen over the last few months, this is starting to make a lot more sense. Half the board was left out of the launch of ChatGPT. The alignment and safety team was disbanded. And we found out that if you worked for OpenAI and decided to leave, you were contractually obligated not to talk bad about OpenAI or potentially risk losing your vested equity. And then you have Sam Altman saying he didn't know that was in there and he feels embarrassed that it was, but he signed all of the paperwork that allowed that to happen. Either he's not being candid about knowing that that was in there, or he's not reading the agreements that he's putting his signature on. And when OpenAI does decide to go and create a new safety board, they actually put Sam and other team members from OpenAI on this safety board, which really feels like it should be an outside counsel kind of situation. You probably want a third party that doesn't stand to profit from the technology. Right? Because if the people that stand to profit from the technology are the ones going, yeah, it's safe, trust us, there's kind of a conflict of interest there. We also know that OpenAI trains on a lot of data, and they're very sort of ambiguous about where that data comes from. And then you have the Scarlett Johansson issue where they're claiming that it wasn't supposed to be Scarlett's voice, but almost all the evidence points to they were trying to replicate what people saw in her, which was Scarlet's voice. Now, I am not a hater of OpenAI. I think a lot of people are going to watch this video and go, is Matt shifting to, like, anti AI and talking crap about OpenAI now? No, that's not what I'm doing. I'm just kind of pointing out all of the facts because I find all this stuff fascinating. I put out a tweet the other day that kind of sums up the way I feel about stuff. Here's the tweet. I don't know who needs to hear this, but it is okay to think a new piece of tech is really cool and also be worried about privacy, security, and the ethics surrounding the creation of that tech. That's the world that I live in. I'm constantly in awe by this tech. I am super impressed by what OpenAI puts out. I think GPT4O was really cool. I think the demos they showed off of what you could do with your camera and the desktop app were really cool. I've made tons of content about ChatGPT and Dall E3 and all the various OpenAI products. I think they're a company that makes really, really cool products. But at the same time, I'm not full on effective accelerationist. Push the boundaries and move everything forward as fast as possible with no concern for anybody else. That's not where I live. I also have concerns about privacy, security, ethics, things like that. And the more of these stories that come out about OpenAI, the more I sort of lose my faith in them as a company. But it doesn't mean I don't think the tech is still really cool, that I'm not impressed by it. But I do have concerns and I think that's okay. I don't think it has to be binary. I think you could live somewhere on this spectrum of I think it's cool, but I'm also kind of concerned about where they're taking things. And everybody who's watching this video is going to fall somewhere in that spectrum one way or the other. And you're likely not going to fall all the way to one side or all the way to the other. You're going to be somewhere in the middle of there, and that is okay. I don't know Sam Altman personally. I really don't know anybody at OpenAI. I don't really have any contacts there other than a few people that I tweet back and forth with sometimes. But I do think most of the people at OpenAI are probably good people with good intentions. The more I learn about Sam, the more I start to wonder, is he sort of this shark wearing a dolphin suit? Whenever you see him speak publicly, he's so eloquent. He chooses his words very carefully. He seems to know what he's talking about, and he seems to really, really care about where all this AI stuff is going and care for the safety of humanity. But then you hear stories from behind the scenes, and this is all stuff that I just find really, really fascinating. I want to see tech progress. I want to see cooler and cooler models come out. I want to see AI create more of a utopia. But at the same time, I often wonder if the people that are trying to get us there are necessarily the right people for the job. And the more these stories come out, the more I question that. And I'm going to continue to keep my finger on the pulse of all this stuff and share the latest news with you. And I'm excited to see what OpenAI does next. I want them to do more really cool stuff. I guess I'm just hoping that maybe we'll get some more transparency. Maybe we'll get a little bit more of the behind the scenes of how this is all working and how it's being trained and what it's being trained on and what their intentions really are. Because right now it's definitely becoming more and more ambiguous and those waters are getting muddied, and I'm just going to leave it at that. I can sort of rant and ramble about what's going on, but it's a really crazy story right now, and I'm going to keep my eye on it and share more of what's going on as I learn about it with you. And if you like videos, this and you want to learn more about AI and what's going on in the world of AI and get some tutorials and learn about cool tools in the AI space, like this video and subscribe to this channel. And I'll make sure more videos like this show up in your YouTube feed. And if you haven't done it Already, check out FutureTools IE and sign up for the free newsletter. I'll keep you looped in on just the most important news that you need to know and the coolest AI tools that I come across. It's totally free to join I'll also send you free access to the AI Income database when you sign up. And when you sign up, you'll also be entered to win in our monthly giveaways. This month I'm giving away an Insta360X4 camera, but every month I'm going to be giving away cool new gadgets and access to software tools and things like that. Just for being a subscriber, you can subscribe by checking out futuretools IO and thank you so much for tuning into my little ramble rant here about what's going on with OpenAI. Hopefully you learned something. Maybe you're seeing it from a different perspective. Maybe you're thinking my view of it is way off base and I either need to be all in or all out. But that's just not how I work. I'm not that binary. My opinions on these types of things are very fluid. I might like one company one day and the next day think the company is the devil. That's just how it works. I don't hold on to opinions too strongly and my viewpoints can be changed all the time based on new information that I learn. And as I learn new information, I've sort of made it my role to try to turn around and share that information with you. So hopefully you appreciate it. And if you do, again, like subscribe, appreciate you. See you in the next video. Bye bye.