Welcome to Volume 2, Issue 3 of “Explore the AI Landscape”.

Today’s issue is packed with some solid content. From newly discovered vulnerabilities in GPU’s to AI now solving complex Geometry and Elon asking for more shares of Tesla, there’s been a plethora of developments in the AI world over the last week.

In the “Tutorial of the Week” section you will find a wonderful nugget of resources you can use (most are free) to enhance your AI/ML skills and knowledge. We highly recommend checking this section out.

Over in “The Ethics Corner”, we present Deloitte’s 2nd Annual “Ethical Technology: Principles for Emerging Tech” which goes through some interesting topics.

In this edition:

Own a tool that you would like to see on our site or newsletter please Submit a tool or reach us by replying to this email or at [email protected]

With the ever-growing sophistication of cyber threats, organizations are turning to cutting-edge technologies to safeguard their digital assets. One such technology that has revolutionized the cybersecurity landscape is Artificial Intelligence (AI). In this blog post, we will explore the top AI-powered cybersecurity solutions, their benefits, and the future trends in this exciting field

Explore the Collection

Visually build and launch native mobile apps with Draftbit

Draftbit is a powerful tool that enables users to visually build native mobile apps, responsive web apps, and internal tools. With a simple interface, it offers a collaborative environment where users can design, build, test, and publish apps, all from the same place.

If you want to share your company or product with fellow AI enthusiasts before we’re fully booked, submit your tool here and sign up!

In the rapidly evolving landscape of artificial intelligence, the demand for powerful computing resources has skyrocketed. Central to this surge is the reliance on Graphics Processing Units (GPUs), which have become indispensable for running Large Language Models (LLMs) and processing vast amounts of data swiftly. This increased dependency on GPUs, however, has unveiled significant security vulnerabilities in chips from major manufacturers like Apple, Qualcomm, and AMD.

Traditionally, GPUs were engineered with a focus on raw graphics processing capabilities, often sidelining data privacy concerns. This approach starkly contrasts with the meticulous security measures integrated into Central Processing Units (CPUs). As GPUs become more integral to AI and machine learning applications, their architectural shortcomings in terms of data security are becoming glaringly evident. Researchers from Trail of Bits have highlighted this issue, pointing out that GPUs could leak substantial amounts of data, ranging from 5 to 180 megabytes, a concerning figure in the realm of CPU security standards.

The vulnerability, termed 'LeftoverLocals', poses a significant threat. It requires the attacker to have some level of operating system access on the target device. Once this hurdle is overcome, the attacker can breach the data silos, accessing and extracting sensitive information from the GPU’s memory. This could include data processed by LLMs, potentially exposing confidential information.

This vulnerability isn't isolated to a single manufacturer; it spans across various GPUs from Apple, AMD, and Qualcomm. Notably, devices like the AMD Radeon RX 7900 XT, Apple’s iPhone 12 Pro, and the M2 MacBook Air are at risk. While Apple has addressed this issue in its latest processors, millions of devices with older chips remain vulnerable. Qualcomm and AMD are also in the process of releasing security updates.

Addressing these vulnerabilities is not straightforward. Even with available patches, the onus falls on device manufacturers to integrate and distribute these fixes to end-users. This multi-layered ecosystem complicates the process of ensuring comprehensive security across all affected devices.

The revelation of GPU vulnerabilities like LeftoverLocals is a wake-up call, emphasizing the need for robust security models in the rapidly growing field of AI and machine learning. As GPUs become more prevalent in various applications, from gaming to healthcare, ensuring the privacy and security of data processed by these chips is paramount.

The integration of CPUs and GPUs in systems-on-a-chip (SoCs) further accentuates the need for enhanced GPU security. As these components share memory resources, vulnerabilities in GPU memory can potentially expose CPU data as well. Moreover, with the increasing trend of GPU virtualization in cloud infrastructure, the risks multiply, making it imperative to establish stringent security protocols for GPUs.

The discovery of the LeftoverLocals vulnerability is a crucial step in understanding and mitigating security risks associated with GPUs. As we continue to push the boundaries of what's possible with AI and machine learning, it's essential to parallelly evolve our approach to data security. This incident should serve as a catalyst for the industry to prioritize and reinforce GPU security, ensuring a safer digital environment for all.

DeepMind has introduced AlphaGeometry, an AI system capable of solving complex geometry problems at a level comparable to a human Olympiad gold-medalist. This breakthrough in AI performance was detailed in a paper published in Nature on January 17, 2024. AlphaGeometry represents a significant advancement in AI reasoning within mathematics, particularly in the field of geometry.

If you’re looking for some great resources around AI and Machine Learning, we recently found this on Github and are excited to dive more into ourselves. In this thread by user duncantmiller, there is a plethora of educational materials, resources and courses from some of the most brilliant minds in the space.

We highly encourage you to take a look for yourself if you have any interest as there is something in there for everyone.

We recently came across this interesting read. It is Deloitte’s second annual report centering around how organizations can proactively promote trustworthy and ethical principles across their orgs and also focus on incorporating ethics as part of the development and implementation of their technological products and services.

If you want to skip some light reding and go straight to the report, it can be found here.

Prompt: a Peach with a large grin going down a water slide

Prompt: A Pig with an Elton John face plays the piano in a bright jacket and sunglasses