Superintelligence in Robots · Part of The Humanoid Group
AGI and Robot AI Safety News
Ambitious talk of superintelligence now sits alongside new safety tools, shifting rules and sober expert forecasts. These figures and dated updates show what changed, with a source for every claim.
SafeWorld raises $12.2 million to test in simulation how robots behave around people
· Funding
SafeWorld, an AI lab co-founded by Ding Zhao, director of Carnegie Mellon's Safe AI Lab, came out of stealth with $12.2 million in seed funding co-led by Shine Capital and a16z Speedrun. Its software simulates rare and dangerous situations to test robots around people, with early pilots at carmakers, warehouse automation companies and medical device makers.
Why it matters: Rare, dangerous situations are hard to test with physical trials alone. Simulation results may become part of the safety evidence buyers expect before robots work near staff, so ask vendors what they have tested.
At the UN, Guterres says killer robots have no place as Trump rejects global AI controls
· Policy
Opening the UN General Assembly debate, Secretary-General António Guterres called for more oversight of AI and said life-and-death decisions must never be surrendered to machines, citing a new UN science report on AI risks. US President Donald Trump rejected any global scheme to control AI and said the technology should now be called super intelligence.
Why it matters: Leaders disagree on whether advanced AI and autonomous machines need global rules. Organisations deploying AI-driven robots across borders should plan for national rules that differ, rather than one international standard.
EU's AI Act overhaul routes AI rules for machines through the Machinery Regulation
· Policy
Regulation (EU) 2026/1744, the Digital Omnibus on AI, was published in the EU's Official Journal. It moves high-risk AI duties for AI built into regulated products to 2 August 2028, and shifts machinery to a separate list, so the AI Act's requirements for machines will arrive through Machinery Regulation delegated acts due by that date.
Why it matters: Makers and users of AI-driven robots in Europe get more time, but the requirements will come through machinery law. Safety files should track the Machinery Regulation and the harmonised standards being written for it.
SoftBank's Masayoshi Son vows to pursue superintelligence and says it now makes robots
· Market
At SoftBank's annual shareholder meeting in Tokyo, chief executive Masayoshi Son, 68, said he will lead the group into his 70s to bring about artificial superintelligence, which he defines as 10,000 times smarter than a human. He said SoftBank has started manufacturing robots at a physical AI plant, with an announcement to follow.
Why it matters: A major AI investor is tying its superintelligence goal to robot manufacturing, which means more capital and bolder claims. Judge vendors on demonstrated performance on your own tasks, not on stated ambitions.
NVIDIA launches Halos for Robotics, a safety system for AI-driven robots
· Launch
NVIDIA announced Halos for Robotics, which pairs its IGX Thor computer and Halos safety software with an inspection lab accredited by the ANSI National Accreditation Board to help partners prepare for third-party certification. Agility is the first adopter and plans to build parts of it into the safety system of its Digit humanoid; the lab counts more than 40 companies.
Why it matters: Certifiable safety layers for robots run by AI models are still scarce. Asking whether a robot's safety functions have been independently assessed, and against which standards, is becoming a practical buying question.
Expert panel's median forecast puts a robot passing the Coffee Test in 2034
· Research
The Forecasting Research Institute's Longitudinal Expert AI Panel published robotics forecasts from 214 experts surveyed in March 2026. Their median forecast has a robot first make coffee without mistakes in three unfamiliar, home-like settings in 2034, and experts expect a robot to perform an appendectomy without physical help around 2040.
Why it matters: Experts see household-level robot skills as years away. Plan near-term deployments around structured, well-defined tasks, and treat claims of human-level robot dexterity with caution.
People also search for superintelligence news, AGI news, AI safety news, robot AI safety news, artificial superintelligence news and AI regulation news.