Chinese AI researchers claim to have developed the first fully autonomous AI agent in history, although some doubt that its potential constitutes another “DeepSeek moment.”

After launching last week, the agent named Manus immediately gained popularity in AI communities.

Its premiere demo went viral, sparking a flurry of online debate. While some praised its potential, others pointed out its shortcomings and raised privacy concerns.

“A Fully Autonomous Agent”

Manus, a chatbot developed by startup Monica, is a subsidiary of Butterfly Effect, a Singapore-registered entity. Unlike traditional chatbots, Manus can handle complex tasks independently, sort resumes, rank candidates, and format data into spreadsheets.

Notably, Yichao 'Peak' Ji, cofounder of Manus, described the autonomous agent as a tool that bridges the gap between conception and execution.

Manus, a cloud-based tool, outperforms OpenAI's Deep Research model on the GAIA benchmark for stock trend analysis, web data scraping, and interactive website creation. Still, access is limited and requires an invitation to test.

The Creation of Manus

Manus, an AI model, has been criticised for not being built from scratch and being more of an AI wrapper running on other models like Anthropic's Claude 3.5 Sonnet. Pierre-Carl Langlais, cofounder of AI startup Pleias, stated that Manus uses token conventions specific to Anthropic.

Moreover, Ji later stated that Manus employs Claude 3.5 Sonnet v1 and refined Alibaba's Qwen models, despite the introduction trailer suggesting that numerous models powered it.

“We use Claude and different Qwen-finetunes,” Ji stated on X Monday, stating the need for numerous auxiliary models, as they initially used Claude 3.5 Sonnet v1 for Manus's development but are currently testing Claude 3.7 internally and expecting updates.

Manus Illustrates Comparison of DeepSeek

Manus and DeepSeek, a Chinese startup, have been compared due to their AI models, which were reportedly built at a fraction of the cost of Western models.

This led to a $600 billion sell-off in Nvidia stock and demonstrated China's ability to build leading AI models despite US export controls restricting access to advanced chips.

Chinese AI startup Manus claims full autonomy, advancing the US AI race. AI policy researcher Dean Ball believes this is not a 'DeepSeek moment', as it surpasses American firms capabilities and becomes the most sophisticated AI computer.

Researchers Vary

AI researchers and founders are divided on the potential of Manus, an agentic AI tool. Some believe it's a breakthrough, while others, according to Pleias founders Alexander Doria and TechCrunch's Kyle Wiggers, Manus was prone to factual inaccuracies, execution issues, and infinite loops during testing.

Conversely, Hugging Face's Victor Mustar called Manus “the most impressive AI tool I've ever tried,” suggesting it could eliminate the need for coding.

AI specialist Langlais accused Manus of using “hunger marketing” strategies and deceptive communication, calling for greater market openness and transparency. However, the functionality review capabilities and UI were commended by him.

Manus: Privacy Worries

AI researcher Luiza Jarovsky has raised concerns about Manus' data storage and access to Chinese authorities, fueled by its opaque ownership.

Corpora.ai CEO Mel Morris sees promise in Manus, but is sceptical due to its ties to China. These factors will intensify the ongoing debate over AI security and governance, as the company's opaque ownership and data privacy concerns could further complicate the situation.

Read more news:

Crude Oil Projections: OPEC+ Summit and US Data in the Spotlight

NASA's Roman Telescope: A Giant Leap Toward Understanding the Universe

Satellites Undergo “Mass Migrations” Due to Geomagnetic Storms