Researchers at UCSB and UCLA published a paper solving the six-year problem of unconditional unclonable encryption. The construction and key proof ideas were entirely generated by an AI agent (Codex on GPT-5.6 Sol Ultra). Two independent proofs using the same model appeared within days.

OpenAI's GPT-5.6 family (Sol, Terra, Luna) is now generally available on Amazon Bedrock, accessible via the OpenAI Responses API on the bedrock-mantle endpoint. The models target different workloads: Sol for deep reasoning and autonomous coding, Terra for balanced production use, and Luna for high-volume low-latency inference. They feature a 272K-token context window, text and image input, and configurable reasoning effort levels.

US lawmakers have introduced a bill that would require AI companies to disable or limit their systems upon order from the Department of Homeland Security. The bill, supported by both Democrats and Republicans, includes triggers such as loss of control causing at least 10 deaths or economic damage over $100 million, with fines up to $20 million per day for violations.

Gigatoken, a BPE tokenizer written in Rust by Stanford PhD student Marcel Rød, achieves text encoding speeds of 24.53 GB/s on a 144-core AMD EPYC system, outperforming HuggingFace tokenizers by 989x and OpenAI's tiktoken by 681x. The library supports 23 tokenizer families and attributes its speed to a hand-optimized pretokenizer using SWAR SIMD techniques and pretoken caching.

Elon Musk praised xAI's Grok 4.5 for disproving a 30-year graph theory hypothesis in just 8 minutes. He also noted that OpenAI's GPT-5 independently solved the same problem, highlighting rapid progress in AI reasoning.

Moonshot AI released Kimi K3, an open-source model claimed to be world-class. The model is designed for general-purpose tasks and is available for public use.

At WAIC 2026, TARS (Tashi Intelligence) unveiled its embodied native foundation model AWE 3.5, capable of multitasking without reloading. The model powers a fully automated wiring harness production line and includes a built-in physics simulator for reinforcement learning in a digital twin, marking a breakthrough in embodied scaling.

The Trump administration is split over how to respond to China's rapid AI advancements, with the White House pushing for stricter controls and the Commerce Department favoring more moderate measures. The debate intensified after Moonshot AI's Kimi K3 model rivaled top US models, and following allegations of distillation attacks by Chinese firms.