Midjourney bans Stability AI: maliciously crawls data, causing server paralysis for 24 hours

Midjourney has blacklisted Stability AI, banning all Stability AI employees from using its software until further notice.

There have been some conflicts between the two AI image generation companies. Although the field of AI mapping is developing rapidly and the market competition is fierce, Midjourney and Stability AI are still popular among users. Even though they are competitors, Midjourney seems to have no reason to ban Stability AI employees from using its software.

According to disclosures from sources, the Midjourney server suffered a severe attack in the early hours of last Saturday, targeting accounts related to Stability AI data collection engineers. These accounts were used to crawl a large number of prompt words and images, causing the Midjourney server to be down for a full 24 hours.

Midjourney engineers flagged the account for the first time in the evening and discovered that it was trying to obtain all prompt words and images for every user member on the site. These requests put tremendous pressure on the company's servers, causing some databases to be paralyzed for up to 24 hours. After banning the account and conducting further investigation, Midjourney employees discovered that the account was using two credit cards linked to a lead data collection engineer on the Stable Diffusion team.

In response, Midjourney has decided to ban all Stability AI employees from using its software until further notice. 》

## Source: https://twitter.com/PirateWires/status/1765484940957823486

## The recent news released by AI enthusiast Nick St. Pierre once again confirmed this. After using Midjourney many times to create, he said, "During Midjourney's working hours, Midjourney employees told me that someone tried to obtain all the information in the middle of the night last Saturday. prompts and images, resulting in their service being paralyzed."

Pierre also released a complete record of service outages and response times, according to the records As can be seen:

Service was disrupted for 24 hours on Saturday due to botnet-style activity on paid accounts;
Midjourney Suspect It was caused by Stability AI employees trying to grab prompt word and image pairs to train an AI model;
decided to ban all Stability AI employees from using the service indefinitely.

But Stability AI CEO Emad said he didn’t know about it, and replied with “Eh what”, saying that he couldn’t understand it either.

In the following reply, Emad continued: "It is confusing how 2 accounts can bring down the service. And we're not scraping the data, we've been using synthetic data and other data (to train the model). Anyway, I'm a big fan of Midjourney and David (Midjourney CEO), which is why I supported them from the beginning and worked for them. The reason for providing funding for the development of the beta version."

"If anyone in the team has really done this (I have already asked, I will continue to investigate, and Midjourney is also welcome to have news It's really not good to contact him directly), but it's clearly not a DDoS attack, but an honest mistake. It's certainly not something Stability AI directed, and we're now very happy with our data set and the improvements we've made based on it."

# Many people expressed disbelief about this matter:

picture

However, serious netizens are beginning to ask for evidence, and hope that Midjourney will release logs or other evidence as soon as possible.

However, some netizens do not take sides. How could a thief be angry because another thief stole something?

The above is the detailed content of Midjourney bans Stability AI: maliciously crawls data, causing server paralysis for 24 hours. For more information, please follow other related articles on the PHP Chinese website!