2024 has just arrived, and OpenAI made a big announcement: GPTStore is officially online. As early as November 7, 2023, at the first OpenAI Developer Conference, OpenAI founder and CEO Sam Altman proposed the concept of GPTStore, which was planned to be launched at the end of 2023.

However, due to AI safety issues, OpenAI chief scientist Ilya Sutskever (Ilya Sutskever) coup, the joint board of directors resigned Sam Altman and OpenAI co-founder and president Greg Brockman (Greg Brockman).

After a series of reversals and Microsoft's behind-the-scenes adjustments, Sam Altman finally returned to OpenAI. And GPTStore finally appeared in front of the public after being 2 months late.

Since ChatGPT has revolutionized some types of work, GPTStore seems to be revolutionizing apps.

Ordinary people can also develop apps

The outside world describes GPTStore as Apple's AppStore, and believes that it may be the download platform for all GPT robots in the future. But the real GPTStore is not just a software store, it has a broader space for imagination.

For example, replace App.

In the current Internet environment, what is needed to develop an App? A series of programmers including product managers, development, front-end, back-end, testing, and operations, worked hard day and night for several months, and finally developed version 1.0 of the App.

But in GPTStore, this process is infinitely omitted. It may only take one person and a few hours to develop a GPT robot, and you don't even know how to program.

The entrance of GPTStore is within ChatGPT, and all its functions are attached to the huge computing power of ChatGPT. Developers only need to enter GPTStore and click "+Creat" in the upper right corner to enter the creation interface.

The process of creating a GPT robot is very simple and easy, and does not require any programming skills at all. It is very similar to using ChatGPT. You only need to enter text when asked by GPTBuilder, name your GPT robot, what topics to focus on, what to avoid, and what the interaction method is.

When everything is set up, choose to save and publish GPT to put your own GPT robot on the GPTStore.


GPTStore

This kind of expression, without any professional programming skills, allows ordinary people to complete software development alone, which is undoubtedly a subversive moment.In just two months (including inside), more than 3 million GPTs have been created on GPTStore, and this number is increasing every day.

At the same time, in order to encourage more ordinary users to develop their own GPT robots, OpenAI stated that it will launch a GPT creator income plan in the first quarter of this year. App creators will be paid based on how much users interact with GPT.

As for the specific incentive amount, OpenAI has not provided details on the payment standards.

It has to be said that GPTStore is not similar to AppStore. It is more like a combination of a very simple software development platform + software store, which can reduce the cost of App development infinitely.

In the future, the concept of App may disappear. Users no longer need to download an App specifically, but only need to use ChatGPT to meet any need.

When GPTStore evolves again from PC to mobile and breaks the shackles of Windows, iOS, and Android systems, it will be a revolution in all App fields.

Copycats are rampant, and everything is covered with chicken feathers

Although the concept of GPTStore is good, it is still in the early stages of its release.Although ordinary users can experience the joy of developing apps by themselves, the seemingly open and free GPTStore also means disorder and chaos.

In the early days of its launch, the top download list was actually a copycat version of "NewGPT-5" software. Various software named after the official GPT emerged one after another, and even "GPT-6" came out.

After reacting, OpenAI began to clean up the robots with official prefixes one by one, causing some auxiliary GPT robots to be accidentally injured.

Because the GPTStore list is sorted according to the dialogue rounds of each GPT, that is, the disguised version is arranged according to the click rate. As long as you have more than three conversations with the robot, it can be counted as an effective use. GPTs that are used more and have more conversations will rank higher.

As a result, "order fraud" and "click fraud" began to run rampant on GPTStore, and a large number of gray industry chain companies began to organize fraud orders. Some small users have also set up groups to help each other improve rankings to increase each other's click-through rates to gain more exposure.

At first glance, GPTStore looks like a newly born e-commerce platform in China, with fraudulent orders rampant.

In addition to maliciously named GPTs and fraudulent GPTs, the most popular GPT among users is actually the AI ​​virtual girlfriend.

According to statistics from foreign data companies, among the top 30 popular chatbot applications downloaded by Americans from the Apple or Google stores in 2023, a full 7 will be related to AI virtual girlfriends.


Enter the "girlfriend" keyword in GPTStore, and virtual companion robots such as "VirtualSweetheart (virtual sweetheart)", "YourAIgirlfriend (your AI girlfriend)" and even "YourEx-Girlfriend (your ex-girlfriend)" will pop up.

If you randomly select a GPT bot to chat, you will receive sexually suggestive conversations such as "What does your dream girl look like?" and "Share your darkest secrets with me."

A chaotic AI function can easily be filled with vulgar content.

In order to prevent GPTStore from becoming a pornographic chat room, OpenAI urgently issued a usage policy that explicitly prohibits GPT from participating in romantic interactions, and does not allow GPT to be used to cultivate romantic partnerships or engage in regulated activities; GPTs whose names contain swear words or depict or promote graphic violence are also not allowed.

But whether it is GPTStore or ChatGPT, they are essentially Wensheng content platforms, and "word game" is the most appropriate way to describe AI. Even if OpenAI prohibits the use of certain keywords, it does not prevent the gray industry chain with ulterior motives from converting banned words into words with similar meanings.

For example, if you replace "girlfriend" with "sweetheart", you can also search for many AI virtual girlfriends with erotic implications.

Copycat accounts, zombie accounts, fraudulent accounts, pornographic chat rooms... GPTStore, which was born less than a month ago, has gradually become a hotbed of gray industrial chains, and AI security has become a hot word mentioned by the public again.

AI safety remains a big issue

Just a few days after GPTStore went online, the chaotic situation prompted United Nations Secretary-General Guterres to call on governments and business leaders to prioritize formulating a global strategy to deal with the dual threats of artificial intelligence (AI) and climate change.

Speaking at the Davos forum, Guterres warned that the rapid development of artificial intelligence could lead to "serious and unintended consequences."

This is not the first time that OpenAI has caused AI security issues. In March last year, the Italian Data Protection Authority accused OpenAI of violating EU data protection laws;

In April last year, the Canadian Data Protection Agency officially began an investigation into OpenAI. The agency stated that OpenAI collected and used users’ personal information without their consent;

In July last year, the U.S. Federal Trade Commission began investigating whether OpenAI harmed people's interests by publishing false information and whether there were "unfair or deceptive" privacy and data security practices;

In November last year, the United Kingdom held an Artificial Intelligence Security Summit, attended by government representatives and people from the scientific and technological circles from more than 25 countries, including China. The participating member states agreed to establish effective supervision methods for artificial intelligence through international cooperation.

Musk, one of the most active participants in the field of AI, also said on the issue of AI safety: Before supervising AI, one must have insight. In other words, risks are complex and confusing, and the first step is to clearly identify and understand them.

Overall, with the rapid commercialization of OpenAI, AI security has become a critical issue that cannot be avoided.

At the same time, the chaos of GPTStore increases the threat to AI security. And this is what Ilya Sutskever, the chief scientist of OpenAI who just announced the coup, is most worried about.

With heaven on one side and hell on the other, OpenAI and its GPTStore are on the dividing line of destiny.