What's New in AI

AheadForm's Origin F1 humanoid robot demonstrates eerily lifelike facial expressions, sparking Westworld comparisons and debate about the future of social robotics.

Caitlin Kalinowski resigned from OpenAI citing concerns about surveillance and lethal autonomy in the company's agreement with the Department of Defense.

Senate Democrats introduce bills to ban autonomous AI weapons and mass surveillance after the Pentagon blacklisted Anthropic. What the legislation says and why it matters.

In a leaked internal letter, Anthropic CEO Dario Amodei condemned OpenAI president Greg Brockman's $25 million donation to a Trump super PAC and accused rivals of 'dictator-style praise.'

Weeks-old rumors of a fundamental AI architecture discovery now point to Anthropic, fueled by the Mythos leak showing capabilities far beyond incremental scaling improvements.

Google hasn't said a word about Gemma 4, and the open-source AI community is growing restless. Prediction markets are open, Reddit is debating, and competitors aren't waiting.

Chinese manufacturers are doubling the VRAM on RTX 4080 SUPERs and selling them for €1,300. The local LLM community is intrigued — and cautious.

Taalas wants to hardwire LLMs directly into silicon. Their ASIC approach delivers 10x Cerebras speed at a fraction of GPU power draw — but each chip runs exactly one model.

A developer used Google's TurboQuant compression in llama.cpp to run Qwen 3.5-9B with 20K context on a base MacBook Air M4 with just 16GB of RAM.

Chinese robotics company Unipath deployed a home robot that cooks, cleans, and makes beds. A video hit 1.2 million views — and one tiny detail stole the show.