There was a thunder on the ground, when a large number of Chinese were enjoying ChatGPT, they were suddenly not allowed to use it!
ChatGPT is blocking accounts on a large scale, especially the accounts of Chinese users, and a large number of users have reported that they have been recruited.
So, why does ChatGPT have to block accounts on a large scale, especially for the Asian region? No official reason has been given yet. Next, we try to analyze what situations may trigger ChatGPT's banning mechanism from the user specification-related documents given by ChatGPT.
ChatGPT banned Chinese accounts on a large scale, probably because many accounts were purchased through third-party agents
Generally speaking, users in mainland China use ChatGPT, and the usual way of operation is: buy an account online, and then go out to use ChatGPT through VPN over the wall. There are two risk points here, one is to circumvent the wall, and the other is to purchase the ChatGPT account of a third-party agent online. The reason why Chinese users are blocked on a large scale requires the answer from the above two links.
Let’s first look at the risks of overcoming the wall.
In mainland China, according to the "Network Security Law of the People's Republic of China" and other relevant regulations, the establishment and use of private networks or proxy services without approval may be punished by law. Specifically, this may include fines, temporary or permanent bans on related services or accounts, and even criminal liability.
Risks that users in mainland China who use VPN or proxy services may face when using ChatGPT include:
Discovered by regulatory authorities and subject to legal sanctions, such as fines, suspension of services or criminal liability; virtual private network or proxy service providers may record users' network activities, which may lead to threats to users' privacy and data security; using VPN or Proxy services may reduce the quality of network connections, resulting in slow and unstable access, thereby affecting user experience.
However, it should be noted that ChatGPT itself has no geographical restrictions. ChatGPT’s user agreement and terms of service are generally applicable to all users, and the risk of being blocked will not be increased just because users are from mainland China.
That is to say, Chinese users use ChatGPT through VPN over the wall, which mainly affects the user experience, such as poor network stability and slow network speed, but it will increase the risk of being blocked.
Since there is no difference in treatment (at least the official document does not clearly state that it should be treated differently), it is a fact that Asia, especially mainland China, has become the hardest-hit area of titles. How can this be explained?
This requires further analysis of the second risk we mentioned above - many users in mainland China do not directly register for official ChatGPT accounts, but directly purchase accounts provided by third-party agents, which is likely to cause users in mainland China to The main reason for the large-scale banning of ChatGPT accounts.
Purchasing a third-party proxy account may violate ChatGPT's user agreement or terms of service. Specifically, the purchase of a ChatGPT account registered by a third-party agent may involve violations of the terms of service and analysis:
Account Transfer: Many service providers' terms of service state that user accounts may not be sold, transferred, or bartered. If ChatGPT's terms of service also contain similar provisions, purchasing and using accounts registered with third-party agents may violate this provision. The relevant terms might be something like: "You agree not to sell, transfer, or barter your account or access to the Services."
Use unofficial channels to obtain services: The terms of service may explicitly require users to obtain and use services only through official channels. Purchasing a third-party proxy account may be regarded as bypassing the official channel, thus violating the relevant regulations of ChatGPT. The relevant terms might be something like: "You agree to obtain and use the Services only through channels provided and authorized by us."
In the case of violation of the ChatGPT terms of service, the service provider may take the following measures: warning, issuing a reminder, requiring the user to stop the violation; temporarily freezing the account, restricting the user's access and use of the account for a period of time; permanently banning the account, permanently restricting the user's access and use account.
In addition, it should be noted that third-party agents may have some malicious violations, which will affect ChatGPT users who purchased accounts from him. Specifically, there are mainly the following situations:
Register accounts with false information: Third-party agents may use false information (such as fake email addresses, false identity information, etc.) to register ChatGPT accounts in batches. This behavior may cause ChatGPT service providers to find that these accounts do not comply with their terms of service, thereby banning these accounts and affecting users who purchased these accounts.
Batch registration using vulnerabilities or automated tools: Third-party agents may use vulnerabilities or automated tools to register ChatGPT accounts in batches. Such behavior may be seen as an abuse of the system, leading to action by the service provider, such as banning the relevant accounts, affecting the users who purchased these accounts.
Account sharing or selling to multiple users: Third-party agents may sell the same ChatGPT account to multiple users. This could lead to account ownership disputes and privacy leaks, affecting users who purchased these accounts. Additionally, service providers may notice unusual login activity and ban related accounts.
Failure to recharge or renew fees as required: Third-party agents may not pay fees to ChatGPT service providers as required, or use improper means to obtain services. This may cause ChatGPT service providers to take measures to ban relevant accounts when problems are discovered, affecting users who purchased these accounts.
Violations using proxy accounts: Third-party agents may use proxy accounts to commit violations, such as publishing illegal information and infringing on the rights and interests of others. This kind of behavior may cause ChatGPT service providers to take ban measures, affecting users who purchased these accounts.
From the above analysis, we can infer the reason why ChatGPT banned Chinese user accounts on a large scale this time (note that it is only a speculation and has not been officially confirmed): ChatGPT does not want to ban accounts in specific regions, but to ban third-party agents Accounts, and many users in Asia, especially mainland China, obtain accounts through third-party agents, which has led to large-scale bans on ChatGPT accounts of Chinese users.
Is there a solution to this problem?
Sorry, no solution!
Due to the special national conditions, Chinese users naturally have various restrictions and risks when using foreign network services. And ChatGPT is just a kind of foreign service, of course, it is also prone to various problems. This risk may be two-way: on the one hand, the Chinese regulators may actively block ChatGPT services, whether it is direct services or "ChatGPT variants" through various agents, the risk of being banned by the regulators is not small; On the one hand, ChatGPT may limit the service provision in China, and maybe this time there is no "regional discrimination" against China, but it does not mean that it will not in the future. In the past two years, large-scale malicious banning of Chinese accounts by foreign Internet platforms such as Facebook and Twitter has been common. It is hard to say whether ChatGPT will treat Chinese users differently in the future, but this possibility exists.
Chinese users use ChatGPT, data security risks are huge
From a deeper level, a large number of Chinese users and even a large number of Chinese AI applications are built on the basis of large foreign models such as ChatGPT. Whether it is an individual user or a country, there are great data security risks.
Personally, ChatGPT’s user data leaks are endless.
For example, on March 25, OpenAI itself confirmed a data security incident - some ChatGPT Plus service subscribers may have leaked some personal privacy and payment information; Italy has become the first western country to ban the advanced chat robot ChatGPT. Italian regulators say they will not only suspend use of OpenAI's chatbot, but also investigate its compliance with data protection regulations (GDPR).
From the perspective of business logic, when users use ChatGPT, data leakage may be caused by the following reasons: network security vulnerabilities, although the developers of ChatGPT will work hard to ensure the security of the system, any online service may be attacked by the network , such as hacker attacks, phishing websites, etc. This may lead to user data being stolen or leaked; third-party service providers leak data, and operators of ChatGPT may cooperate with third-party service providers, such as cloud service providers for data storage and analysis. Although these third parties usually take measures to protect data security, there is still a risk of data leakage; ChatGPT internal personnel leak user data, that is, ChatGPT internal personnel may leak user data due to misoperation or other reasons.
In addition, in some cases, OpenAI or Microsoft, the service provider of ChatGPT, may need to provide user data to US government departments, such as in accordance with legal requirements. If you think this is impossible, just take a look at the previous "Prism Project".
"Project Prism" (PRISM) is a secret intelligence project of the US National Security Agency (NSA), launched in 2007. The project allows the U.S. government to collect user data for intelligence analysis by directly accessing the servers of several U.S. Internet companies. In 2013, former NSA contractor Edward Snowden leaked a trove of classified documents that revealed the existence and workings of Project Prism.
Prism plans to steal data from other countries mainly through the following methods: Direct access to servers, the US government authorized by a secret court to require US Internet companies to provide access to their servers. This allows the U.S. government to access a large number of users' communication records, emails, photos, files and other data; data interception, NSA intercepts and intercepts information from other countries during data transmission through network monitoring equipment.
It should be pointed out that a large number of American technology giants such as Microsoft, Google, Facebook, and Apple have participated in this plan and are in collusion with the US government. On the one hand, according to government authorization, some companies allow the government to directly access their servers and provide user data to the government for intelligence analysis; on the other hand, the US government ensures that these companies cooperate The process is protected by law to avoid legal liability.
Since the United States was able to do this before, it is entirely possible to do so now and in the future. It is foreseeable that the companies behind ChatGPT, whether it is OpenAI or Microsoft, are managed by the US government and obey US laws. If a US intelligence agency wants to obtain ChatGPT user data, it may be possible to obtain it.
More serious than personal data leakage is the potential national data security risks that ChatGPT may bring. To analyze this problem, we need to analyze the data transmission link of a Chinese user using ChatGPT. When a Chinese user bypasses the firewall to use ChatGPT on the US network through a VPN, the data transmission and processing process is roughly as follows:
1. The user establishes a secure connection through VPN: the user first uses VPN software to establish an encrypted tunnel between the local device and the VPN server. This makes the user's internet traffic appear to originate from where the VPN server is located (e.g. the United States) rather than the user's actual location (mainland China).
2. The user device establishes a connection with the ChatGPT server: the user's device establishes a connection with the ChatGPT server through an encrypted VPN tunnel. During this process, the user's request (such as query input) is transmitted through the VPN server and then on to the ChatGPT server.
3. Data transmission to the ChatGPT server: The data uploaded by the user (such as query input) is sent to the ChatGPT server through the VPN tunnel. Since the data is encrypted in the VPN tunnel, it is difficult for third parties to intercept or view the data.
4. The data is processed on the ChatGPT server: After receiving the user's request, the ChatGPT server processes the data according to the user's input and generates a corresponding reply. This processing may take place on servers in the United States or other countries, depending on OpenAI's server deployment.
5. The result is returned to the user: After the processing is completed, the ChatGPT server sends the generated reply back to the user device. Reply data travels through the same VPN tunnel.
6. User equipment receives data: the user's equipment receives the reply data sent by the ChatGPT server through VPN. The data is decrypted through the VPN tunnel, and the replies generated by ChatGPT can be viewed and used by the user.
Through the above data link, it can be found that a Chinese user uses ChatGPT through VPN, and the data is to be uploaded to ChatGPT’s server in the United States. What is more risky than individual users is the uploading of data by users who are corporate employees. This kind of thing has already happened: Recently, Samsung internal employees used ChatGPT to upload the company's internal semiconductor data. According to statistics, many enterprise employees are directly transferring company data to ChatGPT for help in processing, which has also escalated the risk of data leakage for many enterprises.
Just imagine, if a large number of employees of Chinese companies use ChatGPT, it will inevitably lead to a large amount of data transmission to the United States. This is an absolutely intolerable data security risk at the national level.
China's AI ecology must and will inevitably be built on the basis of China's large model
It is also due to data security considerations that even if ChatGPT shows experience, it is impossible to popularize it on a large scale in China, especially the data application systems of enterprise customers cannot be connected to ChatGPT on a large scale.
No matter how good ChatGPT is, it belongs to others after all.
Although Baidu's Wenxinyiyan is far behind ChatGPT, most Chinese companies can only access Baidu's Wenxinyiyan. The actual situation is also the same. At present, there are already tens of thousands of companies queuing up to access Wenxin Yiyan.
I really hope that Baidu's Wenxinyiyan can make rapid progress and narrow the gap with ChatGPT as soon as possible.
For this reason, the author tried ChatGPT and Wenxinyiyan separately today.
Here are two questions.
Question 1: How to realize controllable nuclear fusion, what are the core technologies and methods to realize controllable nuclear fusion, and what are the main difficulties. Then, expand and talk about the current progress in the field of controllable nuclear fusion, and summarize it in the form of a table. The headers are time, organization, and key progress.
The answer given by ChatGPT is as follows:
Immediately afterwards, I asked Wen Xin the same question again, and the answer I got was as follows:
Next, let's take a more challenging question: introduce the core theory and core equations of quantum theory, and give the derivation process of the core theories and equations. In addition, what is the core contradiction between the theory of relativity and quantum theory, and how to reflect this in the equation
No comments:
Post a Comment