Anthropic Fixes Memory, Apple's New Mac, Stability Raises $76M

Anthropic has merged Claude Chat and Claude Cowork's memory systems into a single unified platform accessible across web, desktop, and mobile devices. This addresses a long-standing frustration where information and context were siloed between different platforms. The update includes a memory inspector tool allowing users to read, edit, or delete stored information. By default, sensitive topics like health, religion, politics, and gender identity are excluded from memory storage unless explicitly enabled by the user. Claude will never store government IDs, social security numbers, criminal histories, or immigration status.

Apple released the M6 Mac Mini and M5 Ultra Mac Studio, designed specifically for local AI inference. The M6 is Apple's first two-nanometer chip with a 12-core CPU and up to 40% faster multi-thread performance than the M4, starting at $899 with 16GB memory. The M5 Ultra Mac Studio scales to 512GB unified memory with 1.2TB/second bandwidth, featuring 36 CPU cores and 80 GPU cores, priced at $5,500. These launches follow macOS 26.2 which enabled distributed MLX inference over Thunderbolt 5.

Stability AI raised $76 million in Series B funding from Universal, Sony, Warner, and EA. These investors are strategically chosen partners with whom Stability already has co-development agreements signed in October and November. The company plans to expand its creative production suite and professional services arm, positioning itself as a competitor to Adobe while leveraging its custom-built generative AI models and tools.

Despite predictions by AI pioneer Geoffrey Hinton in 2016 that radiologists would be obsolete by 2021, radiologist headcount has grown by 26% and salaries have increased substantially. The FDA has cleared approximately 1,400 AI-enabled medical devices, with three-quarters targeting radiology. Studies show AI-assisted procedures detect more issues than conventional methods, yet radiologist employment is projected to grow over the next three decades. This demonstrates that AI tools enhance rather than replace human professionals in healthcare.

Taiwan indicted an NVIDIA manager in connection with a $2.5 billion Supermicro server smuggling scheme involving 74 restricted B300 AI servers sent to Chinese customers in violation of U.S. export controls. Nine people were charged with forging documents; an additional 56 servers were intercepted by Taiwanese customs. Supermicro has terminated employees connected to the scheme. Meanwhile, U.S. lawmakers are considering the Remote Access Security Act to address cloud loopholes as Southeast Asian data centers are planned to expand from 2 to 31 locations, potentially allowing Chinese firms to circumvent AI hardware restrictions.