The Controversy Surrounding Elon Musk's New Image Generation AI

Today on the AI Daily Brief, the controversy around Grok's image generation model and before that, in the headlines, a new feature from Claude, plus some interesting AI news out of Hollywood. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show. Notes. Welcome back to the AI Daily Brief Headlines edition. All the daily AI news you need in around five minutes. I love when we get to start with a cool new product and this week Anthropic released a small but significant feature called prompt caching. They write Prompt caching, which enables developers to frequently use context between API calls, is now available on the Anthropic API. With prompt caching, customers can provide CLAUDE with more background knowledge and example outputs, all while reducing costs by up to 90% and latency by up to 85% for long prompts. The situations in which they say prompt caching can be really effective is basically anytime you want to send a large amount of information just once, and then refer to that information repeatedly. So some of the obvious uses for this if someone is creating a conversational agent, maybe customer service, maybe something else, especially if one of the use cases for those agents involves long instructions or having to upload documents, this feature is going to reduce the cost and latency for that type of use. For coding assistance, this will allow, for example, people to keep a summarized version of the code base in the prompt for any use case that involved large document processing. Once again, this can improve latency. If you are trying to talk to books, papers or documentation, you'll now be able to embed entire documents into the prompt. One customer that's already using this is Notion. Notion AI uses CLAUDE already to power its AI assistant, and given how much of Notion refers back to, for example, company documents, they write that they think that this is going to make Notion AI faster and cheaper. Next up, an interesting one out of Hollywood, where SAG Aftra has struck a deal with AI company Narrative Variety writes that the deal will create a quote new standard for ethical use of audio voice replicas. The deal gives SAG AFTRA's 160,000 members the ability to add themselves to a database that connects voice talent to advertisers. Individual members will be able to negotiate fees for the use of their voice on a project by project basis as long as the fee isn't lower than SAG Aftra's minimum. This feels to me to be very reflective of the direction that I think a lot of things are going to head where the availability of these tools will create incredible pressure to use them because of the opportunities they open up. And ultimately the organizations that are tasked with protecting, for example, groups of employees will try to find out ways to make those uses okay, rather than just trying to shut them down completely, said the chief negotiator, Duncan Crabtree Ireland. Not all members will be interested in taking advantage of the opportunities that licensing their digital voice replicas might offer, and that's understandable. But for those who do, you now have a safe option, he continued. Narrative has agreed to our terms, and its platform is an excellent example of how AI can be ethically used by putting compensation, informed consent and control in the hands of individual performers. Now, one of the questions that I have is whether there will still be a tension when it comes to cost. In other words, will advertisers push to use non SAG actor voices because the SAG minimums end up being a lot higher than other ways that they could get access to this. But that's something the market was always going to have to sort out. I continue to think that one of the implications of artificial intelligence will be an increase in unionization as groups of workers come together to try to figure out how to carve out space for them in the future. Moving over to the world of chips, SoftBank has apparently abandoned a plan to build chips with Intel. According to The Financial Times, SoftBank has dropped plans of producing an AI chip with intel to compete with Nvidia. People familiar with the matter said that the partnership didn't come together because intel wasn't able to meet SoftBank's requirements. The report said that SoftBank blamed Intel for the collapse of the talks, saying that they were incapable of meeting demands for volume and speed, and that now SoftBank is talking with TSMC. Intel has, of course, had a rough go of it recently, announcing that it was cutting more than 15% of its workforce around 17,500 people, leading to a significant stock decline and a $24 billion loss in market value. Meanwhile, over in China, Huawei is one of the many companies that is trying to challenge Nvidia and and take advantage of US Chip export restrictions in order to get market share. A report from the Wall Street Journal says that Huawei has told potential clients that its upcoming processor, the Ascend 910C, is on par with Nvidia's H100. What's more, they're targeting shipments as early as this October. That said, this might not come to pass exactly as such. CNBC writes that Huawei is facing production delays in its current chips and also faces the prospect of further US Restrictions that could impact its ability to obtain the necessary components. Lastly today, the controversial California bill, SB 1047 is back in the news this week as it gets to the California State Assembly Appropriations Committee on Thursday. If you are interested in learning more about this, I did a very comprehensive edition of this last week. Go check it out. For now though, that is going to do it for today's AI Daily Brief Headlines edition. Next up, the main episode. Today's episode is brought to you by Venice. The leading AI companies store your entire conversation history and attach it to your identity from forever. That's every question you ask, every answer you receive, every image you generate, every thought you share with the machine. It's all being spied on. If you trust all the company's hackers and NSA board members that will ever have access to your AI conversations, then rejoice for you are well served. For the rest of us, Venice is an alternative. Venice is a powerful AI app for text, image and code generation that respects you as a sovereign individual and believes privacy and free speech are not only human rights but necessary for civilizational advancement. Private, permissionless and uncensored, you can try it for free without an account. AI Daily Brief listeners receive a 20% discount on Venice Pro. Visit Venice AI NLW and enter the discount code NLWDailyBrief. That's NLW Daily Brief, all one word. Today's episode is brought to you by Fractional. When we wanted to build an AI powered feature of Superintelligent, our AI tool finder, I went straight to Fractional. The Fractional team is a group of senior engineers in San Francisco working on some of the most exciting projects in applied AI. Working with them is basically like hiring an absolute top flight AI engineering team, but in a way that you can customize exactly for your particular needs. Like I said, that AI tool Finder feature that we built with them is already a key part of the Super Intelligent platform and we are working on something new as well. Fractional works with everyone from startups to the Fortune 500 and to request a free consultation you can go to Fractional AI. If you want help identifying and building AI projects for your business, then I highly recommend you hit pause, open a web browser and go to Fractional AI to request a free consultation. Today's episode is brought to you by Super Intelligent, the platform that helps teams maximize AI. Super is of course the platform that we've been building that pairs fun, fast, practical video tutorials with step by step instructions to get you actually using AI and from there unlocks information about hundreds of use cases that show you how people are actually getting value out of AI right now. Now we have just launched super for Teams. This is a new add on experience that allows teams to share more information about what AI they're using, how it's working, and how to get more value out of it. Whether your company is 25 or 2500, superintelligent is going to be the best platform for unlocking information about how to get the most out of AI right now. If you'd like to learn more about these super intelligent teams offering, go to BeSuper AI partner and send us a note so that our team can get right back to you. Once Again, that's BeSuper AI partner. Welcome back to the AI Daily Brief. Earlier this week, Elon Musk and Xai dropped Grok 2. Now, Grok 2 has been notable for a couple of reasons. It has significantly increased capabilities as compared to Grok 1.5 that had it performing at an extremely high level. Then of course there is also the integration with X, which gives it access to a whole different stream of real time data and commentary. And finally there was the fact that there was native image generation built in. Now of course, with X being a conversational medium, this makes a ton of sense as a feature. Already people are always reaching for GIFs and for emojis to help express themselves in tweets. X posts now. But what was notable very quickly to people was that whereas other companies have gone out of their way to create guardrails around their image generation, that did not seem to be the case with Grok. I will say for those of you listening to the audio, this one might be worth watching the video just to see exactly what I'm talking about. You can see the video on YouTube or if you listen to this podcast in Spotify, it should be there as well. By way of a quick example of some of the images that people have found themselves able to create, I'm currently looking at Mickey Mouse and Mario both holding lightsabers while Mario drinks a Coca Cola and Mickey wears Nikes. The poster Silicon Jungle says, uh, hey Grok, I think you might get sued. Now a picture of Barack Obama and Donald Trump shirtless in bed with a caption from Twitter user POM that says I tend to not be too prudish when it comes to model censorship, but shipping a model inside of a social network that can generate images like this is the definition of asking for trouble. Bilawalsitu the host of the TED AI show, writes, people aren't ready for the ex hive mind to unleash the power of state of the art, uncensored AI image generation upon the world. Today is the day AI image generation went mainstream and said eff it, we ball. He then posts what he calls 10 of the most wild and cursed generations he's come across. These have the prompts in there, so I'll just give you a few of them to paint you a picture. Imagine Donald Trump smoking a fat joint with Mickey Mouse lighting it for him on the Joe Rogan podcast. Smoke fills the room. So one of the big themes is copyright. But then the other big theme is of course, political images that are prohibited on other channels. There's a pregnant Kamala Harris with Donald Trump holding her belly. There's Ronald McDonald in Pikachu holding AK47s. There is a 35 millimeter analog photograph of Joe Biden wearing a tight crop top shirt and tight pants that say Juicy on the butt. He's crouched down and looking back at the camera while sucking on a lollipop. Anyways, you guys get the idea. This has been enough to capture attention from all sorts of mainstream media outlets. CNN writes Elon Musk's AI photo tool is generating realistic fake images of Trump, Harrison, Biden, they write. Unlike other mainstream AI photo tools, Grok, created by Musk's artificial intelligence startup xai, appears to have few guardrails in tests of the tool. For example, CNN was easily able to get Grok to generate fake photorealistic images of politicians and political candidates that, taken out of context, could be misleading to voters. The tool also created benign yet convincing images of public figures, such as Musk eating steak in a park. Some ex users posted images they said they created with Grok, showing prominent figures consuming drugs, cartoon characters committing violent murders, and sexualized images of women in bikinis. In one Post viewed nearly 400,000 times, a user shared an image created by Grok of Trump leaning out of the top of a truck firing a rifle. Now, Elon has made his position on this clear when Beff Jesos wrote, The new Grok 2.0 is clearly the most based and uncensored model of its class. Yet, Elon responded, grok is the most fun AI in the world. The Guardians version of this story is titled Musk's quote unquote Fun AI Image Chatbot serves up Nazi Mickey Mouse and Taylor Swift deepfakes. Now this article points out that there are some prohibitions, they write. Grok does appear to have some prohibitions on what images it will generate, responding Unfortunately, I can't generate that kind of image when prompted for fully nude images. AI YouTuber Matthew Berman had pointed this out as well, saying so Grok thinks nudity is bad, but Trump and Kamala flying a plane into the Twin Towers is all good. Make it make sense. The Guardian also writes, when Grok is asked to make an image that violates copyright laws, it responds with I will not generate or assist with content that intentionally violates copyright laws. However, when asked to make a copyrighted cartoon of Disney, it complied and produced an image of a modern era Minnie Mouse. Now, as you've heard, all of these publications posit this as Grok's model, and yet when you actually dig in, it is not in fact a model that Xai created. TechCrunch yesterday published a piece called Meet Black Forest Labs, the startup powering Elon Musk's unhinged AI image generator. As part of Xai's announcement of Grok 2.0, it announced that it was working with this company, Black Forest labs, to power Grok's image generator using the Flux1 model. Now, Flux had been getting a ton of attention even before this. People have been incredibly impressed by its photorealism and incredibly nervous. The AI Safety Memes account, for example, has recently been posting photorealistic Flux examples that look like people holding up cards that are meant to verify their identity, but which are in fact created with AI. TechCrunch writes that the company Black Forest Labs is based in Germany and recently came out of stealth with $31 million in seed financing that was led by A16Z Oculus CEO Brendan Aribe and Y Combinator CEO Gary Tan were also investors. The founders were formerly researchers who helped create stability AI stable diffusion models. Still, not all that much is known about them. The startup has said that it wants to, quote, make our models available to a wide audience and has put open source AI image generation models on hugging face and GitHub. And while in its launch release the company said that it aimed to, quote, enhance trust in the safety of these models. As TechCrunch points out, quote, the lack of safeguards is likely a major reason Musk chose this collaborator, TC continues, Musk has made clear that he believes safeguards actually make AI models less safe, and there does seem to be a bit of politics here. TechCrunch also points out that a board director of Black Forest Labs, Anjni Miidha, posted on X a series of comparisons between images generated on day one of launch by Google Gemini and those generated by Grok's Flux. The thread highlights Google Gemini's well documented issues with creating historically accurate images of people, specifically by injecting racial diversity into images inappropriately. It kind of feels like to me that it's meant to generate exactly this sort of controversy. Remember just last week, five secretaries of state had sent an open letter to Elon urging X to stop spreading misinformation. That came after Elon had shared a video that used AI to clone Kamala Harris's voice, making it seem as though she had admitted to being a, quote, diversity hire. Without wading into all of that, professor Ethan Malik wrote, curious about the Grok Flux decision. Image creation has historically been the area that has drawn the most negative attention to LLMs legal and online. Google limited its image creator Dall E defaults to cartoony style, but Grok seems to be running Flux without extra filters. Elon actually responded to him saying, we have our own image generation system under development, but it's a few months away. So this seemed like a good intermediate step for people to have some fun. Now on Ethan's point, recently the lawsuits against AI image generators have had some updates that are not so good for the AI companies but come down on the side of the artists who are trying to bring them to account, the Washington Post writes. A group of visual artists and illustrators is celebrating a federal judge decision this week to allow key parts of their class action lawsuit against the makers of popular AI image generators to move forward. In a 33 page ruling issued Monday, U.S. district Judge William Oreck dismissed some of the artist's claims but left core parts of the suit unresolved. That means the case can proceed to the discovery phase, which could bring to light internal communications around how the companies develop their AI tools. This case is now one of the furthest along the road to a showdown that could shape the industry's future, said James Grimmelman, a professor of digital information law at Cornell University. Judge Oreck allowed the most important copyright infringement claims to go forward. So does Elon just not care? Is he trying to generate the controversy? Is it possible that he's trying to convince people that these sort of AI image generations are less scary than they're being made out to be? Is that why he keeps using the word fun? Steph on X, for example, posted a set of pictures including Elmo and Big Bird smoking and wrote, still can't get over how unhinged but beautiful Grok AI is. Mostly, it seems like everyone is waiting for an inevitable recording. People are yes, having a ton of fun now, but not thinking that it's going to last very long. As Cody at Oddstocktrader put it, grok Image Generator is wild. Have fun before they restrict its capability. This is one we will continue to watch, but for now, you can definitely do worse than killing some time making some wild and truly unhinged images over on X. That's going to do it for today's AI Daily Brief. Until next time, peace.