Amid allegations of data scraping, Midjourney has instituted a company-wide ban on all employees affiliated with Stability AI. The company has attributed a nearly 24-hour service outage to what it describes as 'botnet-like activity' originating from two accounts linked to the creator of Stable Diffusion.
(Photo : STEFANI REYNOLDS/AFP via Getty Images)
Stability AI has unveiled its latest foray into the world of artificial intelligence with the introduction of "Stable Video Diffusion."
Banning Stability AI Employees
Midjourney has taken action against Stability AI by prohibiting its employees from accessing its services. The move comes in response to allegations that staff members from the rival generative AI firm caused a system outage earlier this month while attempting to extract data from Midjourney's platform.
In a communication posted on its Discord server on March 2nd, The Verge reported that Midjourney acknowledged an extended server outage that resulted in generated images not appearing in user galleries.
During a business update call on March 6th, the company attributed the outage to what it described as "botnet-like activity from paid accounts," specifically implicating Stability AI employees.
Service Outage
As recounted by Midjourney user Nick St. Pierre during a call, Midjourney revealed that the service outage occurred due to an attempt by someone from Stability AI to collect all prompt and image pairs on a Saturday night.
St. Pierre noted that Midjourney had identified multiple paid accounts associated with an individual on the Stability AI data team.
In the summary of the business update call held on March 6th, referred to as "office hours" by Midjourney, the company announced an indefinite ban on all Stability AI employees from accessing its service in response to the outage.
Additionally, Midjourney announced a new policy that will prohibit employees of any company engaged in "aggressive automation" or causing service outages from using its platform.
Also Read: Tumblr's Parent Company Enters Agreements with OpenAI, Midjourney for Training Data
St. Pierre raised concerns regarding the allegations to Stability AI CEO Emad Mostaque, who responded on X, stating that he was investigating the matter and refuting any involvement by Stability.
Mostaque expressed confusion over the actions attributed to two accounts, asserting that their team had not engaged in scraping activities as they relied on synthetic and other data, particularly emphasizing the effectiveness of the Stable Diffusion 3 AI model.
He maintained that if any Stability employee inadvertently caused the outage, it was not intentional and certainly not a DDoS attack.
Midjourney's Response
In response to Mostaque's remarks, Midjourney founder David Holz chimed in on the same thread, mentioning that he had forwarded "some information" to aid in Mostaque's internal inquiry. As of now, the situation remains dynamic, with no further updates provided since the exchange on March 6th.
The notion that only two accounts engaging in scraping activities could lead to such a prolonged server outage does raise eyebrows.
Moreover, the irony inherent in this scenario hasn't escaped the notice of online creators, who have vocally condemned both firms for training their algorithms using vast amounts of online data scraped from their creations without permission.
The Verge reported that both Stable Diffusion and Midjourney have faced multiple copyright lawsuits, with the latter specifically accused of establishing an artist database for training its models back in December.
Related Article: Stable Diffusion 3 is Here, New Generative AI Image Tool, Up to 8B Parameters
