Black Forest Labs new image generator, FLUX.1
Grok, the latest release from Elon Musk, unveiled a new AI image-generation feature on Tuesday night. Similar to the AI chatbot, this feature lacks sufficient safeguards. That implies you have the ability to create fabricated images of Donald Trump appearing to smoke marijuana on the Joe Rogan show, for instance, and easily share them on the X platform. However, it is important to note that the driving force behind the frenzy is not Elon Musk’s AI company but rather a new startup called Black Forest Labs.
The partnership between the two was unveiled when xAI announced its collaboration with Black Forest Labs to enhance Grok’s image generator using the FLUX.1 model. Black Forest Labs, an AI image and video startup that recently launched on August 1, seems to align with Musk’s vision for Grok as an “anti-woke chatbot,” offering a more flexible approach compared to OpenAI’s Dall-E or Google’s Imagen. The social media site is already filled with shocking images from the new feature. 
Black Forest Labs, a Germany-based company, has emerged from stealth mode with an impressive $31 million in seed funding. According to a press release, Andreessen Horowitz was in charge of the funding round. Notable investors also include Garry Tan, the CEO of Y Combinator, and Brendan Iribe, the former CEO of Oculus. The co-founders of the startup, Robin Rombach, Patrick Esser, and Andreas Blattmann, have a background in research and were instrumental in developing Stability AI’s Stable Diffusion models.
According to Artificial Analysis’s analysis, the FLUX.1 models created by Black Forest Lab outperform the Midjourney and OpenAI AI image generators in terms of quality. This conclusion is based on the rankings provided by users in the image arena.
The startup claims to be making their models accessible to a broad audience by providing open-source AI image-generation models on platforms like Hugging Face and GitHub. The company has announced its intention to develop a text-to-video model in the near future.
According to the company’s launch release, their goal is to “increase confidence in the safety of these models.”. However, critics argue that the overwhelming number of AI-generated images released on Wednesday had the opposite effect. Using Google or OpenAI’s image generators, it was impossible to reproduce a number of images that users created using Grok and Black Forest Labs’ tools, including Pikachu holding an assault rifle. It is evident that copyrighted imagery was utilised for the model’s training.
That is precisely the purpose.
It is highly probable that the absence of safeguards played a significant role in Musk’s decision to select this collaborator. Musk has expressed his belief that safeguards can potentially compromise the safety of AI models. “Training AI to be socially conscious, or, in other words, promoting dishonesty, can have serious consequences,” Musk warned in a tweet from 2022.
Anjney Midha, a board director at Black Forest Labs, recently shared a series of image comparisons on X. These comparisons showcased the images generated on the first day of launch by Google Gemini and Grok’s Flux collaboration. The thread brings attention to the well-documented issues of Google Gemini when it comes to creating historically accurate images of people. Specifically, it raises concerns about the inappropriate injection of racial diversity into these images.
“I appreciate @ibab and the team for addressing this matter appropriately,” Midha tweeted, highlighting FLUX.1’s apparent disregard for the issue (and mentioning xAI lead researcher Igor Babuschkin’s account).
As a result of this mistake, Google issued an apology and disabled Gemini’s capability to generate images of individuals in February. Currently, Gemini is still unable to generate images of individuals, as per the company’s policy.
An overwhelming deluge of false information
The absence of proper safeguards could potentially pose challenges for Musk. After explicit, artificial intelligence-generated images of Taylor Swift spread on the platform, the X platform came under fire. In addition to that incident, Grok consistently produces misleading headlines that frequently appear to users on X.
Last week, five secretaries of state appealed to X to cease disseminating false information about Kamala Harris on X. In a recent incident, Musk shared a video that utilised AI to replicate Harris’ voice, creating the illusion that the vice president had confessed to being hired solely for the sake of diversity.
Musk appears determined to allow such misinformation to spread on the platform. By enabling users to share Grok’s AI images without any watermarks, he has essentially unleashed a flood of misinformation onto everyone’s newsfeed.
