Tokyo-based startup Integral AI, founded by ex-Google veteran Jad Tarifi, has announced what it calls the world’s first Artificial General Intelligence (AGI) system, representing a potential leap beyond current AI technologies. The company claims its model can learn new tasks autonomously, without pre-existing datasets or human intervention, signaling a shift toward human-level reasoning in AI.

Integral AI defines AGI using three core principles: autonomous skill learning, safe and reliable mastery, and energy-efficient acquisition. The system is designed to teach itself entirely new skills in novel domains, operate safely without unintended consequences, and consume energy roughly equivalent to a human learning the same skill. According to the company, these benchmarks guided the development and testing of its first-of-its-kind AGI learning system.

The startup has conducted preliminary robot trials, where machines reportedly acquired new skills independently. CEO Jad Tarifi described the achievement as “more than a technical milestone, it marks the next chapter in human civilization,” with ambitions to scale the model toward embodied superintelligence that enhances freedom and collective agency.

While skepticism remains regarding claims of the first true AGI, Integral AI asserts its system mirrors the human neocortex, responsible for conscious thought, perception, and language, positioning it as a potential fundamental leap in AI innovation.