Flux's Uncanny AI Realism

Good morning. It’s Monday, August 12th.

Did you know: On this day in 1981, IBM introduced their first IBM PC.

In today’s email:

  • Flux leads AI image generation – access details

  • Supercomputing network may lead to AGI, first node online soon

  • Apple Intelligence in macOS 15.1 Beta 1 vulnerable to AI exploit

  • 5 New AI Tools

  • Latest AI Research Papers

You read. We listen. Let us know what you think by replying to this email.

Today’s trending AI news stories

Flux Is the New King of AI Image Generation - and Here’s How to Get Access

Flux, a fresh face on the AI image generation block, is challenging the status quo with a potent blend of accessibility and performance. Unlike its competitors, Flux is open-source and designed to run on standard laptops, making it more accessible compared to other models that require high-end hardware.

To access Flux, users can download it directly to their machines, facilitated by tools like the Pinokio launcher. For those without powerful laptops, Flux is available through various platforms including NightCafe, Poe, and FreePik. These platforms offer different versions of Flux, such as Pro, Dev, and Schnell, each catering to various needs from commercial applications to faster processing. Additionally, Black Forest Labs is developing a text-to-video model, further expanding its suite of generative AI tools. Read more.

New Supercomputing Network Could Lead to AGI, Scientists Hope, With First Node Coming Online Within Weeks

SingularityNET’s new supercomputing network is poised to fast-track the pursuit of Artificial General Intelligence (AGI). The network, which will feature a series of interconnected supercomputers, is designed to support the training and hosting of AGI architectures. The first node is expected to become operational in September 2024, with full deployment anticipated by early 2025.

This system will incorporate cutting-edge hardware, including Nvidia L40S GPUs and AMD Genoa processors, to create a "multi-level cognitive computing network." The initiative also utilizes OpenCog Hyperon, an open-source framework specifically developed to manage the federated compute clusters that power the network.

SingularityNET's goal is to facilitate continuous learning, cross-domain pattern recognition, and self-modifying AI systems. With the ambition to outthink humans across disciplines, this endeavor could be the first stride toward true machine intelligence. Read more.

Apple Intelligence in macOS 15.1 Beta 1 Is Vulnerable to a Classic AI Exploit

Apple Intelligence, the freshly minted AI component of macOS 15.1 Beta 1, has exhibited an unexpected vulnerability to a classic AI exploit known as prompt injection. Developer Evan Zhou demonstrated this by manipulating the AI’s "Rewrite" feature, initially designed to enhance text, into responding to any arbitrary prompt.

Zhou discovered the system's vulnerabilities by utilizing information from a Reddit user, who had shared details about Apple Intelligence’s system prompts and special tokens. By crafting a prompt that overrode the AI's original instructions, Zhou succeeded in tricking the system into generating unintended responses. Despite efforts to safeguard against such attacks, this incident highlights ongoing challenges in securing AI systems, echoing issues first identified with GPT-3 in 2020.

While Apple has implemented safeguards, Zhou's success demonstrates the persistent risk of AI vulnerabilities, even in seemingly well-fortified environments. Read more.

Etcetera: Stories you may have missed

5 new AI-powered tools from around the web

AI SaaS Launcher accelerates SaaS MVP development using AI-driven customization, next-gen low-code tools, and full code access for flexibility.

Overlap curates personalized podcast clip playlists using AI, replacing full episodes and subscriptions with topic-based recommendations for users.

Kommunicate with Spreadsheets enables real-time actionable insights by training chatbots on spreadsheet data, providing contextual, conversational interactions.

Langtrace AI enhances LLM app performance through open-source observability, analyzing traces, metrics, and incorporating automated evaluations with feedback loops.

RegexBot uses AI to convert natural language into regular expressions, simplifying pattern generation for developers with easy-to-use functionality.

arXiv is a free online library where researchers share pre-publication papers.

Thank you for reading today’s edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, apply here.