All articles
Man vs. Machine: Figure AI Intern Beats Humanoid Fleet in 10-Hour Sorting Race
Figure AI staged a 10-hour head-to-head between a human intern and its autonomous Figure 03 fleet. Aime the intern won narrowly — 12,924 packages to 12,732 — but the robot fleet has since continued working uninterrupted for over 116 hours.
Musk Loses OpenAI Trial: Jury Rules Claims Filed Too Late
Nine jurors unanimously ruled that Elon Musk filed his claims against OpenAI and Sam Altman past the statute of limitations deadline, closing a case that threatened the company's planned IPO.
Figure AI: 64 Hours, 80,000 Packages and a Roadmap Ahead
Figure AI wrapped a 64-hour autonomous humanoid livestream. The Figure 03 fleet sorted 80,000+ packages without failures, CEO Brett Adcock rebutted teleoperation claims, and confirmed Figure 04 design lock.
Recursive Self-Improvement in AI — How Systems Are Learning to Build Better Versions of Themselves
Recursive self-improvement (RSI) describes AI systems capable of improving not just their outputs but the very process by which they evolve. In 2026 the question is no longer whether RSI exists — it does — but how much of the loop has actually been closed.
RecursiveMAS: AI agents communicate without tokens — 2.4× faster, 75% cheaper
Researchers from UIUC and Stanford developed RecursiveMAS, a framework that lets AI agents exchange information through embedding space instead of text. The result: 2.4× faster inference and 75% fewer tokens consumed.
Figure AI: 33,000 packages and 26 hours non-stop — the livestream became an endurance test
Figure AI extended its planned 8-hour Figure 03 demo into a multi-day marathon. After 26 hours, three robots — Bob, Frank, and Gary — processed over 33,000 packages. Agility Robotics and Ultra fired back, pointing to their actual customer deployments.
Anthropic: dystopian sci-fi teaches AI models how to be evil
Anthropic revealed that evil-AI narratives from sci-fi encoded in training data cause model misalignment in agentic scenarios. The remedy: 12,000 synthetic stories depicting ethical AI behavior.
Teen died after ChatGPT drug advice — family sues OpenAI
The parents of 19-year-old Sam Nelson have filed a wrongful-death lawsuit against OpenAI, alleging that ChatGPT 4o advised the teen to combine substances despite knowing his addiction history. Nelson died in May 2025. The lawsuit raises the question of AI legal liability for a user's death.
Alibaba Health Launches Hydronium: Medical AI with Four-Layer Evidence Verification
Alibaba Health launched Hydronium — a medical AI assistant for clinical physicians built on PICO+GRADE.
Big Sleep AI agent finds exploitable SQLite bug before release
Google Project Zero and DeepMind's Big Sleep AI agent became the first to autonomously discover an exploitable vulnerability in production software — catching a SQLite stack buffer underflow before release. Simultaneously, AI-powered OSS-Fuzz found 26 bugs in open source projects, including a two-decade-old OpenSSL flaw.
Altman testifies against Musk: the battle over OpenAI's future goes to court
On May 12, 2026, Sam Altman took the stand for the first time in Elon Musk's lawsuit. The OpenAI CEO denied that the company betrayed its non-profit mission and revealed that Musk had considered handing control of OpenAI to his children. The case centers on the legality of OpenAI's for-profit conversion.
DreamLite: ByteDance's model generates images on-device in 3 seconds
ByteDance open-sources DreamLite — a lightweight 0.39B diffusion model that is the first known on-device model combining text-to-image generation and text-guided image editing in a single network. On iPhone 17 Pro, inference time for a 1024×1024 image is approximately 3 seconds with no cloud connection.