The Evolution and Security of AI: From Manus's Breakthrough to the Application of FHE

The field of artificial intelligence has recently witnessed a significant breakthrough. An AI system named Manus achieved state-of-the-art results in the GAIA benchmark test, outperforming large language models of its kind. Manus has demonstrated remarkable capabilities, being able to independently handle complex tasks such as multinational business negotiations, which involve contract clause analysis, strategy formulation, and proposal generation, among other aspects.

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

However, the emergence of Manus has also sparked discussions in the industry about the development path of AI: should it move towards a single system of Artificial General Intelligence (AGI), or a collaborative model of Multi-Agent Systems (MAS)? This question actually reflects the balance between efficiency and safety in AI development. As individual AI systems come closer to AGI, their decision-making processes also become increasingly opaque; while Multi-Agent Systems can disperse risks, they may miss critical decision-making opportunities due to communication delays.

The progress of Manus has also amplified the inherent risks in AI development. For example, in medical scenarios, AI systems need to access sensitive patient data; in financial negotiations, undisclosed corporate financial information may be involved. Additionally, AI systems may exhibit algorithmic bias, such as discrimination against certain groups during the hiring process. There is also the potential for adversarial attacks, where hackers may disrupt the judgment of AI systems through specific means.

These challenges highlight a key issue: the more intelligent AI systems become, the wider their potential attack surface.

To address these security challenges, several solutions have been proposed in the field of cryptography:

1. Zero Trust Security Model: This model is based on the principle of "never trust, always verify," and rigorously authenticates and authorizes every access request.

2. Decentralized Identity (DID): This is an identity recognition standard that does not require centralized registration, providing important support for the Web3 ecosystem.

3. Fully Homomorphic Encryption (FHE): This is an advanced encryption technology that allows computation on data in an encrypted state, enabling data processing while protecting privacy.

Among them, fully homomorphic encryption is considered a key technology for solving security issues in the AI era. It can play a role in the following aspects:

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

• Algorithm level: Achieving "encrypted model training" through FHE, so that even developers cannot directly understand the AI's decision-making process.

• Collaborative level: In multi-agent systems, threshold encryption is used for communication, so that even if a single node is compromised, it will not lead to global data leakage.

Although Web3 security technologies may not have a direct connection to ordinary users, they have an indirect impact on everyone. In this challenging digital world, continuously strengthening security measures is a necessary means to protect one's interests.

As AI technology continues to converge with human intelligence, non-traditional defense systems are becoming increasingly important. Fully Homomorphic Encryption (FHE) not only addresses current security issues but also prepares for a more powerful AI era in the future. On the road to AGI, FHE is no longer an option but a necessity to ensure the secure development of AI.