Agility Robotics' Senior Director of Robotic Safety, Kevin Reese, says safety is a single word, and that is exactly what makes it easy to underestimate. Building it into a humanoid robot, he argues, ranks among the hardest problems in robotics. The company released the discussion as a full episode.
Roboticist Aaron Ames says safety is the crucial missing element before humanoid robots can take on larger roles in society. He warns that dazzling demonstrations of jumping, dancing, and kickboxing should not distract from the need for rigorous safety design.
US Vice President JD Vance said many Americans' first reaction to AI or "SI" is to compare it to Skynet, but he dismissed fears of Terminator-style robots taking over nuclear weapons as impossible.
NVIDIA unveiled its Open Agent Safety Platform, built on the open-source OpenShell runtime and the Sentry watchdog, to govern robots running AI agents. Gecko Robotics is among more than 100 organizations testing the framework, applying enforceable boundaries to its climbing, crawling, flying and swimming inspection robots, including the U.S. Navy's Komodo.
Tech leaders and lawmakers express concerns that Asimov's Three Laws are insufficient to address modern AI risks. Calls for slowing AI development and new legislation grow as experts warn of potential harm to society.
Agility CTO Pras Velagapudi says a robot working in a real facility sees data that must be protected, including faces, address labels and proprietary IP. Once a humanoid leaves the controlled environment, data security can no longer be an afterthought, he argues.
UBTech released a Humanoid Robot Technology Ethics Governance White Paper, proposing a full-process framework for ethical risk management. It introduces three protection goals—controllable action, appropriate anthropomorphism and bounded interaction—and a lifecycle governance model with nine stages.
UN Secretary-General António Guterres used his final General Assembly address to warn that rapidly advancing AI could shift power from governments to a few private companies and individuals, and he urged international agreement on new safety safeguards.
Agility Robotics thanked MarketWatch for looking past the hype and pointed followers to its article on factory robots, which have long been seen as a safety threat to human workers and which one company says it has solved.
Researchers at EPFL propose a new discipline called 'Sustainability Robotics', which evaluates robots by their overall contribution to sustainability goals. They argue that robots must be minimally invasive, accessible, and symbiotic, and that sustainability should become a key performance metric.
NEURA Robotics has written an article on data sovereignty in cognitive robotics and invited readers to take a look.
NEURA Robotics published an article on data sovereignty in cognitive robotics, following its post asking whether companies would share process data to train robots.
Physical AI introduces new robot safety risks beyond component failure, because attackers can manipulate perception, decisions, and actions. Research across model backdoors, system exploits, and runtime manipulation shows cybersecurity must join functional safety throughout the robot lifecycle.
Agility Robotics' Digit 5 is programmed to avoid obstructing humans, a first for humanoids. With a graded safety motion system, 65,000 operational hours, and OSHA recognition, it targets the trust needed for factory deployment.
Daniel Diez, CBO at Agility, says the public fixation on home robots misses the key requirement: cooperative safety, meaning a robot can work in close proximity to people. He argues the path to the home runs through industry, where the technology must mature first. The remarks follow Agility's unveiling of Digit 5, a humanoid designed to work safely alongside humans, as reported by Bloomberg.
Police in East Kazakhstan deployed robot dogs for real patrol duties at the Nomad Alemi nomadic culture festival, marking the region’s first operational use of robotic systems for public order. The robots offer mobile video surveillance, facial recognition, and loudspeaker warnings, raising questions about data retention and the scope of their role.
Business Insider asked ChatGPT, Gemini, Claude, and Grok to rank paths to human extinction. The models agreed that killer robots are the least likely threat; the real danger is humans using AI for bioweapons and cyberattacks.
Agility published a look at how the safety systems on its Digit humanoid have evolved across generations. The company said the safety architecture improves through real-world experience, and the piece covers lessons learned from deploying Digit in industrial environments.
On September 7, about 30 robots staged a 'protest' in Warsaw, Poland, organized by Democratism to discuss AI's impact on jobs. The event, however, was a PR stunt by a robotics company CIO, highlighting the pressing issue of accountability when AI systems cause harm.
Unitree's G1 humanoid robot was shown fighting a human in a fully autonomous demo, but independent analysis suggests the AI may run on an external computer. The real concern is not the fight itself, but the lack of safety frameworks for increasingly capable humanoid robots after a G1 kicked a child at an amusement park.
In an April 2026 security demonstration, a commercial quadruped robot was hijacked remotely and turned on a human dummy, prompting 360 founder Zhou Hongyi to warn that embodied AI converts code flaws into physical destruction. Research by DARKNAVY and others found a well-known humanoid robot could be fully compromised in under eight hours, and the Unitree G1's "UniBLEed" flaw chain may even be wormable, while standards and security investment race to catch up.
At IFA 2026, combat robots drew attention, prompting CNET's Katie Collins to question whether they normalize violence. Her concerns highlight safety gaps and the need to stop glorifying robot fights.
Zhou Hongyi, founder of 360 Group, urged the cybersecurity industry to quickly build AI-driven vulnerability discovery, automated defense, and intelligent agent security frameworks to address the changed attack-defense landscape.
Robots are increasingly used for environmental tasks, but their own environmental footprint remains under-examined. New projects and tools aim to quantify and improve robot sustainability across the full lifecycle.
In a TechCrunch piece, Agility Robotics CEO Peggy Johnson said robot safety can't be retrofitted, calling it a redesign. She highlighted the safety bar between lab demos and real deployment. The company is preparing to go public, though Johnson isn't promising a robot in your home anytime soon.