Has China just built the world’s first human-level AI?

Manus can buy property, book holidays and create video games, writes Anthony Cuthbertson. But experts warn that releasing it could cause chaos

'Han the Robot' on stage before a discussion about the future of humanity and artificial intelligence (AI) at the RISE Technology Conference in Hong Kong on 12 July, 2017 (AFP via Getty Images)

Just over a month after China experienced its “Sputnik moment” for AI with the release of DeepSeek, another artificial intelligence announcement has provoked speculation that researchers may have made the ultimate breakthrough.

The creators of Manus, which launched last Friday, claim it is the “world’s first” fully autonomous AI agent, capable of performing complex tasks like booking holidays, buying property or creating podcasts – all without any human guidance.

Yichao Ji, who led the development of the AI, said Manus marked the next evolution of artificial intelligence, and offered “a glimpse” into artificial general intelligence (AGI): AI that rivals or surpasses human intelligence.

“This isn’t just another chatbot or workflow, it’s a truly autonomous agent that bridges the gap between conception and execution,” he said in a video demonstrating the AI’s capabilities. “Where other AI stops at generating ideas, Manus delivers results. We see it as the next paradigm of human-machine collaboration.”

Within days of launching, invitation codes to join early testers of Manus were reportedly being listed on online marketplaces for 50,000 yuan (£5,300).

The hype is rivalling that of OpenAI’s ChatGPT, which became the fastest-growing app in history following its launch in November 2022, which itself raised the question of AGI after outperforming humans in everything from law exams to generating code.

ChatGPT and other advanced AI models that have since launched from the likes of Google and Anthropic have disrupted the previous consensus that AGI might arrive within the next decade or two, with some figures within the AI industry believing it may be much closer.

Last month, OpenAI boss Sam Altman said that artificial general intelligence was “coming into view”, which could prove to be a pivotal moment in human history.

“The future will be coming at us in a way that is impossible to ignore, and the long-term changes to our society and economy will be huge,” he wrote in a lengthy blog post.

Anthropic chief executive Dario Amodei, whose company produced the ChatGPT rival Claude, predicts that AGI could come about as early as 2026.

In a 15,000-word essay published in October 2024, Amodei wrote that AGI – or “powerful AI”, as he refers to it – will likely arrive far sooner than most people think due to the rapid development of large language model (LLM) AI systems.

In Amodei’s forecast, the AI will be “smarter than a Nobel Prize winner” and capable of carrying out tasks in a similar way to those outlined by Manus AI.

“It can engage in any actions, communications, or remote operations,” he wrote, “including taking actions on the internet, taking or giving directions to humans, ordering materials, directing experiments, watching videos, making videos, and so on. It does all of these tasks with a skill exceeding that of the most capable humans in the world.”

Similar capabilities are listed on Manus’s website, however after just a few days of testing, some users have observed the AI making errors that most humans would spot.

In an analysis of the gaming console market, it only included reports for the Sony Playstation and Microsoft Xbox.

“Even for one of the biggest companies in the world with countless reports online, Manus makes easy to spot mistakes, missing the Nintendo Switch entirely,” one X user noted.

Manus said it was currently working to resolve any problems before a wider release, using the limited roll out to identify potential issues.

“The primary goal of the current closed beta is to stress-test various parts of the system and identify issues,” a spokesperson said. “We deeply appreciate the valuable insights shared by everyone.”

It will take more rigorous testing to prove whether Manus can actually be considered AGI, though it is already reshaping the debate over what is considered human-level artificial intelligence, in the same way that ChatGPT forced researchers to find better methods to assess AI than the decades-old Turing test.

These early flaws, however, have led experts to warn that AI agents like Manus should not yet be allowed to take over human tasks without proper oversight.

“If given autonomy over high-stakes tasks—such as buying and selling stocks—such imperfections could lead to chaos,” Mel Morris, chief executive of AI-driven research engine Corpora.ai, told The Independent.

“Vigilance in deployment, instrumentation, and monitoring is critical. Unfortunately, little progress has been made in these areas, which must be urgently addressed. We need to understand what our AI agents are doing, how they reach their conclusions, and establish mechanisms for oversight or intervention.”