The Future of Privacy in AI and Cryptocurrency

In yesterday's article, I shared the risks that AI applications may pose to personal privacy.

The reason for this risk lies in the operational model of current AI applications—when end users call upon AI, they directly upload their data to the AI large model located in the cloud, which then infers user behavior characteristics based on that data.

The longer this data accumulation lasts, the more comprehensive the AI large model can understand user behavior habits through algorithms.

For individuals, this risk translates to privacy leakage, while for companies, it becomes a leakage of trade secrets.

In an interview video last year, Jensen Huang mentioned this risk regarding AI and strictly required his employees to know what data must remain local and what data can be uploaded to the cloud when using AI tools.

However, at that time, I only heard this risk as information, and it was only when I personally experienced this risk that I recalled his warning.

This issue has only just begun to surface, but it will soon escalate.

Therefore, I believe that in the future, as AI becomes more widespread, we will need a terminal (whether a phone, glasses, or some other form we cannot yet imagine) that runs a streamlined large model locally, processes most sensitive data locally, infers simple requests, and only uploads some heavy tasks and data to the cloud after "filtering" for the complex large model to handle. This way, it avoids the cloud's large model from directly capturing users' personal behavior characteristics.

In the crypto ecosystem, privacy handling has long been on the agenda.

Earlier, Vitalik mentioned that public chains like Ethereum hindered their widespread use in the commercial sector due to the transparency and openness of data and information—because in many cases, parties in commercial transactions are unwilling to disclose transaction information to protect trade secrets.

Recently, some commercial users pointed out that they are currently quite cautious about the large-scale use of stablecoins. This is because stablecoin accounts on public chains are all public, meaning anyone can see how much stablecoin any account holds. Once the identity information of an account is leaked, the amount of funds (stablecoins) held by any company/enterprise becomes transparent information.

Therefore, whether in AI or the crypto ecosystem, privacy issues are problems that must be solved next.

However, if we carefully compare the applications of AI and the crypto ecosystem in terms of privacy, it seems that crypto applications are currently ahead—there have long been privacy coins (like Monero, ZCASH, etc.) and mixers in the crypto ecosystem.

But these privacy applications actively avoid regulation, so they have been somewhat labeled negatively.

The privacy applications that can be accepted by the public and regulators now seem to be more mature and feasible, possibly based on zero-knowledge proof solutions, such as a method that has been experimented with:

Having a licensed regulatory institution act as an intermediary, with both parties conducting transactions through this intermediary, while concealing identities and transaction information, only placing the final generated zero-knowledge proof on the public chain for certification.

This way, transaction information ensures privacy, and both parties avoid criminal suspicion.

I hope that in terms of privacy protection and transaction compliance, the exploration and application within the crypto ecosystem can leverage its advantages to first explore a new path for AI applications.