We're excited to support BitRobot in open-sourcing the largest humanoid whole-body teleoperation dataset collected in real homes. We hope it accelerates progress toward general-purpose humanoid robots.😉
Wuji Tech has unveiled the Wuji Hand 2, a robotic hand with fast and human-like finger movements.
Built with metal components, gears, and compact electric motors, it offers high precision and control.
Advanced sensors automatically adjust grip strength, allowing it to handle different objects securely and smoothly.
Wow. This is huge.
Researchers from UC Berkeley, NVIDIA and Stanford introduced T-Rex, a framework that combines vision, language and touch so robots can react to physical contact in real time instead of relying on vision alone.
This system is trained on a 100-hour tactile-synchronized dataset spanning 200+ everyday objects and 22 motor primitives, teaching robots how to feel and manipulate the physical world.
Across 12 contact-rich manipulation tasks, T-Rex achieved over 30% higher average success rates than the strongest baseline models.
BREAKING: The NSA's own director says Mythos broke into almost all of its classified systems in hours.
Per The Economist, Senator Mark Warner, vice chair of the Senate Intelligence Committee, said General Joshua Rudd, who runs the NSA and the Pentagon's Cyber Command, told him this directly.
This came out on June 11, the same day Amazon reportedly found a separate jailbreak in Anthropic's models. Within hours, Trump ordered Anthropic to cut off foreign access to Mythos and Fable.
Anthropic shut both down completely instead.
Now there are two competing stories for why this actually happened.
One says the shutdown was a response to the NSA's own classified systems getting breached in hours.
The other says Anthropic is privately pushing back, calling the jailbreak minor and the shutdown an overreaction to something other AI models can already be tricked into doing.
The NSA was already using Mythos for its own cyber operations, with Anthropic engineers embedded inside the agency. The same tool the agency was actively relying on is the one its own director says broke into almost everything it owns.
Alex Karp was asked the question every Palantir investor has been dreading, can the large language model companies just replicate what Palantir does? (Save this).
His answer was that the investors asking that question are reading reports about something they do not understand technically.
The LLM companies solve the simplest, easiest problems that sell tokens and they fundamentally do not understand how disliked they are outside of San Francisco.
There is a real product argument underneath that.
When you need to put a missile on an adversary's head and bring Americans home safely, you do not hand that task to a probabilistic model built to be right 51% of the time.
Manufacturing a car part, launching a rocket, running battlefield intelligence, these are problems where being wrong is catastrophic, and no serious enterprise would trust that to a system designed to generate convincing text.
He also dropped a line that most people missed.
Most of what Anthropic talks about in public is actually running on Palantir.
Dario Amodei builds the frontier model, Karp likes him personally but the deployment layer underneath is Palantir's.
The frontier model and the operating system are two different businesses, and Karp is betting the next seven years belong to whoever controls the operating system.
The bull case is not subtle at this point.
Palantir just reported 85% revenue growth in Q1 2026, its fastest expansion since going public with US revenue up 104% year-over-year and US commercial revenue up 133%.
They posted a Rule of 40 score of 145%, raised full-year 2026 guidance to $7.65 billion implying 71% growth, and the US business is on pace to double again in 2027.
The reason Palantir is hard to replicate is exactly what Karp described, it is not a software product, but rather an embedded operating system for the world's most complex organizations.
Switching costs are enormous, trust takes years to build, and the cultural gap between Palantir's defense-first DNA and a San Francisco LLM startup is not something you close by hiring good engineers.
That is not a company being disrupted by large language models but rather the company that everything else runs on top of.
Come join Milk Road Pro for our full breakdown on Palantir, what the Anthropic partnership means for long-term revenue, and our entire AI thesis.
Link below!
ロボットの『目』に小さなシールを貼るだけで動作を妨害できる、現実的な攻撃手法が学術的に報告されました。安全認証を考えるうえで無視できない問題提起です。
ロボット用 AI モデル(VLA)に対し、攻撃側が機体の中身を知らず限られた動作軌跡データだけを持っている前提でも成立する『敵対的パッチ攻撃』が体系化されました。手順は二段階で、まず AI が画像のどこに注目しているかを推定し、次にその部分に物体認識を狂わせるパッチを貼って動作軌道を歪めるよう最適化します。シミュレーションと実機の両方で、作業の成功率が大幅に低下することが示されました。
ヒューマノイドが工場・店舗・公共空間で動く前提に立つと、安全規格や認証の議論に『敵対的攻撃への耐性』を入れる必要性が現実味を帯びます。