CMU

OpenAI
2026-09-20 04:41:22

GPT-6 Astra’s robot demos revive a familiar question: could OpenAI become the OpenAI of robotics?

A low-cost SO-101 robotic arm drawing the Golden Gate Bridge under GPT-6 Astra’s control has pushed OpenAI back into the center of robotics discussion. The demo was not based on a preset path: Astra planned the strokes, watched the canvas through a camera, and adjusted as it drew. Since then, a series of public experiments have shown the model handling one-shot task imitation, building real2sim environments from robot demonstrations, and controlling more complex embodiments in simulation and on real hardware. The strongest data point came from RoboCurve, which connected GPT-6 Astra to two real I2RT YAM robotic arms. In a pick-and-place test, Astra succeeded 19 times out of 20, or 95%, while Fable 5.1 succeeded 8 times in the same setup. Those results have sharpened a broader industry debate: if a general-purpose model is already strong enough at perception, reasoning, and action planning, robotics may not need to train every “brain” from scratch on massive robot-specific datasets. The article also traces OpenAI’s earlier robotics work with Dactyl, the company’s retreat from the field because of data constraints, and its possible return through a different route. At the same time, Astra still shows clear limits in millisecond-level control and in the “last millimeter” of precise insertion tasks, where contact, force, and hardware error remain hard problems.

170
GPT-6 Astra’s robot demos revive a familiar question: could OpenAI become the OpenAI of robotics?
Hong Kong
2026-09-19 07:19:55

Hong Kong to pilot tokenized Exchange Fund Bills before end-2026

Hong Kong will pilot the tokenization of Exchange Fund Bills before the end of 2026, Secretary for Financial Services and the Treasury Christopher Hui said at a press briefing on financial development measures tied to the city’s first five-year plan for economic and social development and the Chief Executive’s 2026 Policy Address. Hui said digital assets and fintech are new growth engines for Hong Kong, adding that digital bonds issued in the city account for about 50% of the global total. He also said the Central Moneymarkets Unit, or CMU, will build a digital asset platform during 2026 to provide one-stop services for digital bond issuance and settlement. On regulation and market development, Hong Kong plans to refine its virtual asset licensing regime, improve the regulatory framework for tokenized investment products, support trading of regulated stablecoins on licensed platforms, and encourage broader use cases for compliant stablecoins.

170
Hong Kong to pilot tokenized Exchange Fund Bills before end-2026
Hong Kong
2026-09-19 07:19:43

Hong Kong targets tokenized Exchange Fund Bills pilot by end-2026

Hong Kong Financial Secretary for Financial Services and the Treasury Christopher Hui said digital assets and financial technology are serving as new growth engines for the city, outlining a set of policy and market-development steps tied to the government’s financial agenda. Speaking at a press conference on financial development measures under Hong Kong’s first five-year plan for economic and social development and the Chief Executive’s 2026 Policy Address, Hui said digital bonds issued in Hong Kong account for about 50% of the global total. He also said Hong Kong plans to pilot the tokenization of Exchange Fund Bills before the end of 2026. The Central Moneymarkets Unit, or CMU, is set to build a digital asset platform within 2026 to provide one-stop services covering digital bond issuance and settlement. On the regulatory side, the city will work on improving its virtual asset licensing regime and the oversight framework for tokenized investment products, while also promoting trading of regulated stablecoins on licensed platforms and encouraging broader use cases for compliant stablecoins.

110
Hong Kong targets tokenized Exchange Fund Bills pilot by end-2026
Axis Robotics
2026-08-01 06:26:40

Axis Robotics raises $12 million seed round as browser-based robot data network tests its POINTS model

Axis Robotics said on July 27 that it had raised a $12 million seed round led by Hack VC, with participation from Nomad Capital, Pi Network Ventures and 10K Ventures. The company positions itself as infrastructure for physical AI rather than a model or hardware builder, using browser-based teleoperation to let users collect robot interaction trajectories without owning robots. Its system pairs remote simulation tasks with a Web3-based attribution layer on Base. Each approved trajectory receives a Data ID tied to a contributor wallet, submission time, quality score and task metadata, while the full data stays off-chain. Axis says this structure is meant to turn contributor output into verifiable records that can later support reward distribution, licensing or governance. The incentive stack now includes POINTS, possible fiat revenue sharing from enterprise task packages, and a token that has not yet launched. The company has disclosed community experiments, academic benchmarks and a list of partners, but key business details such as revenue-sharing ratios, contract values and repayment data remain undisclosed. That leaves a central question unresolved: whether enterprise demand can become recurring enough to support the broader incentive loop.

2370
Axis Robotics raises $12 million seed round as browser-based robot data network tests its POINTS model
World Models
2026-07-06 06:44:26

MemoBench Exposes a Core Weakness in World Models as 10 Video Generators Score Below 0.6 on Object Reappearance

Researchers from Harvard, MIT, IBM, Boston University, Google, Johns Hopkins, CMU, and the Kempner Institute have introduced MemoBench, a new benchmark designed to test whether video generation models can preserve object permanence in dynamically changing environments. The benchmark addresses a major blind spot in current world-model evaluation: most existing tests focus on frame consistency while objects remain visible, but rarely measure whether a model can maintain identity, update state, and restore an object correctly after it leaves the camera view and continues changing off-screen. Built on 360 high-quality ground-truth videos spanning both synthetic and real-world scenes, MemoBench evaluates 10 leading video and world-generation models using automated metrics and VQA-based semantic scoring. The headline result is that no model achieved an object reappearance score above 0.6 out of 1. The findings suggest that visually coherent generation still falls far short of genuine world understanding, especially when memory continuity and state evolution under occlusion are required.

610
MemoBench Exposes a Core Weakness in World Models as 10 Video Generators Score Below 0.6 on Object Reappearance