Apple's head of AI resigns after Siri problems

A recently leaked internal document, referred to as the "Soul Doc," has revealed how Anthropic programs the personality and ethical guidelines of its AI model, Claude 4.5 Opus. The authenticity of this document has been confirmed by Anthropic, indicating a unique approach to AI character development in the industry.

All major web browsers now support WebGPU, following updates from Mozilla and Apple, enabling desktop-class graphics on Windows, Mac, and Android devices. This new API allows browsers to access a device's GPU for running complex applications that were previously limited by the older WebGL standard.

John Giannandrea, Apple's head of artificial intelligence, has announced his retirement effective in 2026, raising concerns about potential shifts in the company's AI strategy. This change comes amid ongoing challenges with Apple's Siri voice assistant, which has faced criticism for its performance.

Nvidia showcased new AI models for autonomous driving and speech processing at the NeurIPS conference, highlighting its commitment to advancing artificial intelligence technologies. This presentation underscores Nvidia's role as a leader in AI innovation, particularly in sectors that require sophisticated machine learning capabilities.

OpenAI is set to launch a new reasoning model next week, which internal tests suggest will outperform Google's Gemini 3. This release indicates a strategic shift within OpenAI, prioritizing this model over other ongoing projects.

Apple has appointed a former Google executive as its new vice president of artificial intelligence, signaling a strategic move to bolster its AI capabilities. This hire comes as Apple seeks to enhance its technological offerings in a competitive landscape dominated by AI advancements.

Kling AI has launched Video O1, touted as the world's first unified multimodal video model capable of both generating and editing video content within a single framework. This innovative approach aims to streamline video production processes, making it easier for creators to manage their projects efficiently.

The ARC benchmark, once a formidable test of AI systems' fluid intelligence, has recently shown signs of decline as modern AI optimization techniques continue to advance, challenging its status as a reliable measure of cognitive capabilities.