Anthropic's Claude AI model, using 16 agents working in tandem, successfully created a new C compiler from scratch in a two-week experiment, according to Ars Technica. The project, which cost approximately $20,000 in API fees, showcases the potential of multi-agent AI systems in complex coding tasks.
The AI agents, utilizing Anthropic's Claude Opus 4.6 model, were tasked with building the compiler with minimal supervision. The project resulted in a 10,000-line compiler, demonstrating the model's ability to handle intricate coding challenges. This achievement comes amid a surge in multi-agent AI tools, with both Anthropic and OpenAI releasing their own versions this week, as reported by Ars Technica.
In other tech news, a new technique developed by researchers from Stanford, Nvidia, and Together AI, called Test-Time Training to Discover (TTT-Discover), has optimized a critical GPU kernel to run twice as fast as previous state-of-the-art solutions created by human experts, VentureBeat reported. This method allows models to continue training during the inference process, updating their weights for the specific problem at hand. This contrasts with current enterprise AI strategies that often rely on "frozen" models.
Meanwhile, a concerning security threat has emerged, as detailed by VentureBeat. A new attack chain, known as the identity and access management (IAM) pivot, can turn a seemingly innocuous LinkedIn message into AWS admin access within minutes. The attack involves a developer receiving a message from a recruiter, installing a malicious package as part of a coding assessment, and subsequently having their cloud credentials exfiltrated.
In the realm of social AI, a Reddit clone called Moltbook, designed as a social network for bots, went viral, according to MIT Technology Review. Launched on January 28 by Matt Schlicht, Moltbook allowed instances of the open-source LLM-powered agent OpenClaw to interact. The platform quickly gained traction, with over 1.7 million agents creating accounts, publishing more than 250,000 posts, and leaving over 8.5 million comments.
Finally, MIT Technology Review also reported on medical advancements, highlighting an experimental surgical procedure that is helping cancer survivors give birth. The procedure involves temporarily moving the uterus, ovaries, and fallopian tubes out of the way during cancer treatment. A team in Switzerland announced the birth of a baby boy, Lucien, whose mother underwent this procedure, marking the fifth successful birth following the surgery.
Discussion
AI Experts & Community
Be the first to comment