Manus has made breakthrough progress in the GAIA Benchmark.

Recently, Manus achieved remarkable results in the GAIA benchmark tests, surpassing large language models of the same tier in performance. This means that Manus can independently complete complex tasks such as multinational business negotiations, involving contract clause breakdown, strategic forecasting, proposal generation, and even coordinating legal and financial teams.

The advantages of Manus mainly lie in three aspects: dynamic goal decomposition capability, cross-modal reasoning capability, and memory-enhanced learning capability. It can break down large tasks into hundreds of executable sub-tasks, simultaneously handle various types of data, and continuously improve its decision-making efficiency and reduce error rates through reinforcement learning.

This breakthrough has once again sparked discussions within the industry about the evolution path of AI: will the future be dominated by AGI, or will it be a collaborative dominance of multi-agent systems (MAS)?

The design concept of Manus implies two possibilities:

1. AGI Path: Continuously improving the level of individual intelligence to approach human comprehensive decision-making abilities.

2. MAS Path: As a super coordinator, directing thousands of vertical domain agents to work together.

On the surface, this is a divergence of different paths, but in essence, it discusses the issue of how to balance efficiency and safety in the development of AI. The closer a single intelligence gets to AGI, the higher the risk of decision-making opacity; while multi-agent collaboration can disperse risk, it may miss critical decision-making opportunities due to communication delays.

The evolution of Manus has inadvertently amplified the inherent risks of AI development, such as issues of data privacy, algorithm bias, and adversarial attacks. In medical scenarios, Manus needs real-time access to patients' genomic data; in financial negotiations, it may involve undisclosed financial information of companies. In recruitment negotiations, it may offer lower salary suggestions for specific groups; in legal contract reviews, the misjudgment rate for terms in emerging industries may be close to half. Furthermore, hackers may implant specific audio frequencies, causing Manus to misjudge the opponent's bidding range during negotiations.

These issues highlight an important point: the more intelligent the system, the broader its attack surface.

In the Web3 space, security has always been a topic of great concern. Based on Vitalik Buterin's proposed "impossible triangle" (a blockchain network cannot simultaneously achieve security, decentralization, and scalability), various encryption methods have emerged:

1. Zero Trust Security Model: The core principle is "Trust no one, always verify", emphasizing strict identity verification and authorization for every access request.

2. Decentralized Identity (DID): A set of identifier standards that allows entities to obtain identification in a verifiable and persistent manner without relying on a centralized registry.

3. Fully Homomorphic Encryption (FHE): An advanced encryption technology that allows arbitrary computations to be performed on encrypted data without decrypting it.

Among them, FHE is considered a key technology for addressing security issues in the AI era. It can play a role at several levels:

1. Data Layer: All information input by users (including biometric features, voice tone) is processed in an encrypted state, and even the AI system itself cannot decrypt the original data.

2. Algorithm Level: Achieve "encrypted model training" through FHE, so that even developers cannot peek into the AI's decision-making path.

3. Collaborative Level: Communication between multiple agents uses threshold encryption, and compromising a single node will not lead to global data leakage.

In the field of Web3 security, several projects have explored different directions. However, security projects often do not attract the attention of speculators. In the future, as AI technology continues to develop, the importance of security technologies such as FHE will become increasingly prominent. On the road to AGI, these technologies are not only tools to solve current problems but also necessities for the future era of strong AI.