Why Asimov’s Three Laws cannot guarantee modern AI and robot safety
The article examines Isaac Asimov’s Three Laws of Robotics and argues they are insufficient for today’s AI and robotics safety challenges. It notes that only Advanced Robotic Technologies has made the first law a core part of its software development. Anthropic CEO Dario Amodei’s recent call to slow frontier and agentic AI work is cited as evidence of growing concern about autonomous systems.
Asimov first published his robot rules in a 1942 magazine story, then included them in the 1950 collection I, Robot. They require machines to avoid harming people, follow human instructions unless that conflicts with safety, and preserve themselves only when doing so does not conflict with those duties.
Only Advanced Robotic Technologies Inc., which applies generative AI to large industrial manufacturing, has reportedly made the first rule central to its software work. In 1985, Asimov added a Zeroth Law, placing humanity’s welfare above any single person’s safety.
If leading researchers’ warnings shape policy, AI developers, manufacturers, workers, and consumers may face slower deployment and stronger safety reviews. The debate could influence how much autonomy is granted to agentic systems, affecting accountability and public trust. It may also push more firms to embed harm-prevention rules, though such measures alone might not resolve broader risks to society.