Anthropic's Claude AI model, utilizing sixteen agents, successfully created a new C compiler from scratch in a two-week experiment, according to a recent blog post. The project, which cost approximately $20,000 in API fees, showcases the potential of multi-agent AI systems in complex coding tasks.
The experiment, detailed in a blog post by Anthropic researcher Nicholas Carlini, involved sixteen instances of the company's Claude Opus 4.6 AI model. These agents were tasked with building the compiler with minimal supervision, working on a shared codebase. The AI agents reportedly produced a 10,000-line compiler. This comes as both Anthropic and OpenAI are releasing multi-agent tools.
In other news, defense attorneys are seeking access to investigative files related to the killing of Renee Nicole Good, after learning that the ICE agent who shot and killed her was the same officer who was assaulted by Roberto Carlos Muñoz-Guatemala. Attorneys for Muñoz-Guatemala, convicted in December of assaulting ICE officer Jonathan Ross, requested a federal judge order prosecutors to turn over training records and investigative files related to the Good shooting, which occurred on January 7 during Operation Metro Surge.
Meanwhile, a new attack chain is emerging that can compromise cloud environments through seemingly legitimate LinkedIn messages. According to a CrowdStrike Intelligence research report published on January 29, the identity and access management (IAM) pivot allows attackers to gain access to cloud environments within minutes. The attack involves a developer receiving a LinkedIn message from a recruiter, installing a malicious package for a coding assessment, and having their cloud credentials exfiltrated.
In the realm of AI advancements, researchers from Stanford, Nvidia, and Together AI have developed a new technique, TTT-Discover, that optimizes GPU kernels. This technique allows a model to continue training during the inference process, updating its weights for the specific problem. TTT-Discover managed to optimize a critical GPU kernel to run twice as fast as previous state-of-the-art solutions created by human experts.
Finally, Valve announced that the rising costs of components like RAM and storage may impact the pricing and shipping schedule of its Steam Machine. The company wrote in a blog post that it must revisit its pricing strategy for the living room-focused PC gaming box. Analysts were divided on how much of the increased component costs Valve would pass on to consumers.
Discussion
AI Experts & Community
Be the first to comment