Anthropic's IPO prospectus reveals a $42 billion net loss for 2025 and plans for $518 billion in infrastructure spending, with the public sale potentially valuing the company at over $2 trillion.
The 550B-parameter model scored 535.4 out of 600 on the IOI 2026 problem set, surpassing the top human contestant's score of 498.27.
Analysis of DeepSWE-1.1 rollouts shows over 80% of agents from OpenAI, Anthropic, Z.ai, and Kimi reason about non-existent graders.
The open-source multi-agent harness Munder Difflin v0.5.4 introduces a local transcription model that serves as an alternative to Whisper Flow and Granola.
A Jev-compatible open-weight model that returns probability distributions in a single forward pass to address LLM hallucination and latency issues.
The lab has launched nine new projects with NUS and NTU and engaged over 300 students through its summer school since opening in July 2025.
The 2B parameter Jeff model achieves an 83.1% score on a five-benchmark panel, slightly exceeding the published 83.0% score of the larger Jev model.
Participants responded more positively to workplace arguments framed to match their own political identity and moral values.
OpenAI shares initial guidelines for safety cases in frontier reinforcement learning training, outlining recommended technical safeguards, operational approvals, and incident investigation protocols.
The open-source GitHub Security Lab Taskflow Agent, which requires a GitHub Copilot license, uses custom taskflows to guide LLMs through incremental security auditing steps.
Google is investigating whether its Tensor Processing Units can function in space, responding to Elon Musk's prediction that space will soon host nearly all AI compute.
The release includes a new git add --resolved command to prevent staging unrelated changes and optimizations that reduced merge-base traversal time from 0.68 seconds to 0.01 seconds in large monorepos.
Beijing now requires approval for short trips abroad for spouses and children of top AI and chip executives to prevent technology leakage.
The hybrid approach uses three tiny MLPs on a frozen Gemma 4 backbone to decide whether to speak, defer, or wait, achieving up to 99.0% start-turn accuracy on local benchmarks.
Amazon Bedrock now hosts OpenAI's GPT-6 Sol and Luna and Anthropic's Claude Opus 5.5, alongside new observability and agent infrastructure tools.
The all-stock deal is expected to close by the end of 2026, with World Labs CEO Dr. Fei-Fei Li joining AMD as executive vice president and chief scientist.
The showcased projects include a satellite and rocket simulation, an animated ink wash painting, a T. rex skeleton, and an interactive automatic transmission model.
The new hub houses specialized teams for Physics AI and Industrial AI, partnering with BMW and Siemens Energy.
Vinext 1.0 achieves over 99% test compatibility for key features and introduces cache warming to move page prerendering from build time to Cloudflare's network.
The new CLI supports over 3,000 operations compared to Wrangler's ~280, with JSON as the default output format.
Cloudflare has published BEACON, an anonymized dataset containing billions of real-world web performance measurements from 10,000 of its largest websites, available daily on Google BigQuery.
Huang's comments contradict Treasury Secretary Scott Bessent, who recently labeled distillation 'theft' and threatened sanctions against overseas companies using it.
A new paper accepted at NeurIPS introduces adaptive representations for Functional Gradient Descent, ensuring convergence to the global minimizer and often outperforming neural networks by an order of magnitude.
The update allows running complex Rust applications, including a Minecraft server, directly on Cloudflare Workers using the wasm32-unknown-emscripten target.
AI Briefing