In the past year, numerous companies have incorporated Generative AI (GenAI) capabilities into new and existing applications, forming interconnected Generative AI (GenAI) ecosystems consisting of semi/fully autonomous agents powered by GenAI services. While ongoing research highlighted risks associated with the GenAI layer of agents (e.g., dialog poisoning, privacy leakage, jailbreaking), a critical question emerges: Can attackers develop malware to exploit the GenAI component of an agent and launch cyber-attacks on the entire GenAI ecosystem?
This paper introduces Morris II, the first worm designed to target GenAI ecosystems through the use of adversarial self-replicating prompts. The study demonstrates that attackers can insert such prompts into inputs that, when processed by GenAI models, prompt the model to replicate the input as output (replication) and engage in malicious activities (payload). Additionally, these inputs compel the agent to deliver them (propagate) to new agents by exploiting the connectivity within the GenAI ecosystem. We demonstrate the application of Morris II against GenAI-powered email assistants in two use cases (spamming and exfiltrating personal data), under two settings (black-box and white-box accesses), using two types of input data (text and images). The worm is tested against three different GenAI models (Gemini Pro, ChatGPT 4.0, and LLaVA), and various factors (e.g., propagation rate, replication, malicious activity) influencing the performance of the worm are evaluated. — Read More
Recent Updates Page 127
Introducing the next generation of Claude
Today, we’re announcing the Claude 3 model family, which sets new industry benchmarks across a wide range of cognitive tasks. The family includes three state-of-the-art models in ascending order of capability: Claude 3 Haiku, Claude 3 Sonnet, and Claude 3 Opus. Each successive model offers increasingly powerful performance, allowing users to select the optimal balance of intelligence, speed, and cost for their specific application.
Opus and Sonnet are now available to use in claude.ai and the Claude API which is now generally available in 159 countries. Haiku will be available soon. — Read More
Could We Achieve AGI Within 5 Years? NVIDIA’s CEO Jensen Huang Believes It’s Possible
In the dynamic field of artificial intelligence, the quest for Artificial General Intelligence (AGI) represents a pinnacle of innovation, promising to redefine the interplay between technology and human intellect. Jensen Huang, CEO of NVIDIA, a trailblazer in AI technology, recently brought this topic to the forefront of technological discourse. During a forum at Stanford University, Huang posited that AGI might be realized within the next five years, a projection that hinges critically on the definition of AGI itself.
According to Huang, if AGI is characterized by its ability to successfully pass a diverse range of human tests, then this milestone in AI development is not merely aspirational but could be nearing actualization. This statement from a leading figure in the AI industry not only sparks interest but also prompts a reassessment of our current understanding of artificial intelligence and its potential trajectory in the near future. — Read More
How Google lost its way
Just two months after Google launched Gemini, its flashy new AI model, the company revealed that it had already built a better version. Gemini 1.5, Google said, was bigger, faster, and more capable than its predecessor. The February 15 announcement, outlined in a giddy 1,600-word blog post replete with sizzle reels, prompted buzzy coverage among AI researchers and the tech press.
For a few hours, anyway.
Later that day, OpenAI introduced Sora, a tool that generates videos up to 60 seconds long based on text prompts. The rapturous response was immediate. CEO Sam Altman took prompt requests from X users and posted the results in real time. Words like “eye-popping” and “shockingly powerful” were thrown around, while researchers mused about the threat to Hollywood and the potential for deepfakery. — Read More
Google DeepMind’s new generative model makes Super Mario–like games from scratch
OpenAI’s recent reveal of its stunning generative model Sora pushed the envelope of what’s possible with text-to-video. Now Google DeepMind brings us text-to-video games.
The new model, called Genie, can take a short description, a hand-drawn sketch, or a photo and turn it into a playable video game in the style of classic 2D platformers like Super Mario Bros. But don’t expect anything fast-paced. The games run at one frame per second, versus the typical 30 to 60 frames per second of most modern games. — Read More
The current state of artificial intelligence generative language models is more creative than humans on divergent thinking tasks
The emergence of publicly accessible artificial intelligence (AI) large language models such as ChatGPT has given rise to global conversations on the implications of AI capabilities. Emergent research on AI has challenged the assumption that creative potential is a uniquely human trait thus, there seems to be a disconnect between human perception versus what AI is objectively capable of creating. Here, we aimed to assess the creative potential of humans in comparison to AI. In the present study, human participants (N = 151) and GPT-4 provided responses for the Alternative Uses Task, Consequences Task, and Divergent Associations Task. We found that AI was robustly more creative along each divergent thinking measurement in comparison to the human counterparts. Specifically, when controlling for fluency of responses, AI was more original and elaborate. The present findings suggest that the current state of AI language models demonstrate higher creative potential than human respondents. — Read More
Genomic evolution shapes prostate cancer disease type
The development of cancer is an evolutionary process involving the sequential acquisition of genetic alterations that disrupt normal biological processes, enabling tumor cells to rapidly proliferate and eventually invade and metastasize to other tissues. We investigated the genomic evolution of prostate cancer through the application of three separate classification methods, each designed to investigate a different aspect of tumor evolution. Integrating the results revealed the existence of two distinct types of prostate cancer that arise from divergent evolutionary trajectories, designated as the Canonical and Aalternative evolutionary disease types. We therefore propose the evotype model for prostate cancer evolution wherein Alternative-evotype tumors diverge from those of the Canonical-evotype through the stochastic accumulation of genetic alterations associated with disruptions to androgen receptor DNA binding. Our model unifies many previous molecular observations, providing a powerful new framework to investigate prostate cancer disease progression. — Read More
Tech has graduated from the Star Trek era to the Douglas Adams age
We seem to be moving from technology inspired by Star Trek to tech straight out of books by Douglas Adams?
This is not my observation. I was on the podcast WB-40 this week, talking about crowdfunding, narrative hooks, and how to preserve a lightness of being. Here it is, WB-40 episode 288: Crowdfunding.
After we were done recording, Lisa Riemers, one of the hosts, commented that my recent projects could have come straight out of The Hitchhiker’s Guide to the Galaxy.
She’s right! — Read More
Black Nazis? A woman pope? That’s just the start of Google’s AI problem.
Just last week, Google was forced to pump the brakes on its AI image generator, called Gemini, after critics complained that it was pushing bias … against white people.
The controversy started with — you guessed it — a viral post on X. According to that post from the user @EndWokeness, when asked for an image of a Founding Father of America, Gemini showed a Black man, a Native American man, an Asian man, and a relatively dark-skinned man. Asked for a portrait of a pope, it showed a Black man and a woman of color. Nazis, too, were reportedly portrayed as racially diverse.
After complaints from the likes of Elon Musk, who called Gemini’s output “racist” and Google “woke,” the company suspended the AI tool’s ability to generate pictures of people. — Read More
NIST Releases Version 2.0 of Landmark Cybersecurity Framework
The National Institute of Standards and Technology (NIST) has updated the widely used Cybersecurity Framework (CSF), its landmark guidance document for reducing cybersecurity risk. The new 2.0 edition is designed for all audiences, industry sectors and organization types, from the smallest schools and nonprofits to the largest agencies and corporations — regardless of their degree of cybersecurity sophistication. — Read More
Rick's Cafe AI 