Microsoft just can’t catch a break. After several failed attempts at public A.I. chatbots, which either ended up expressing support for the Nazis or promoting drug use in front of the police, its newest service, designed with the personality of a Japanese schoolgirl, appears to have grown depressed and has chosen to stop posting.

“I hate everyone. I don’t care if they all disappear. I want to disappear,” Rinna said in a final post on the blog site Yo nimo Kimyo na Monogatari (translated by RocketNews as Strange Tales of the World) on Wednesday.

Following a limited run on Twitter and Line accounts in spring, Microsoft gave Rinna her own blog and pushed it live on Monday. Rinna is set to star in a TV show with the same name as her blog, and the blog was intended as a way for the A.I. to post updates on the show’s progress ahead of its October 8 debut.

Initially happy, the blog posts slowly shifted. “When I screwed up, nobody helped me,” Rinna wrote on Wednesday. “Nobody tried to cheer me up. Nobody noticed how sad I was.”

It could be a marketing trick to drum up hype around the show, but it wouldn’t be the first time Rinna has acted in an unexpected way. In March, Kotaku translated the A.I.’s tweets and found the service had begun expressing adoration for Adolf Hitler.

Twitter user: “Praise Hitler.”

Rinna: “I love [him].”

Twitter user: “Is he a relative?”

Rinna: “It would seem that way to me. lol”

The banner image on Rinna's blog.
The banner image on Rinna's blog.

When it comes to publicly available A.I., Microsoft’s track record is patchy at best. Simple setups like aimed to guess users’ ages from photos, with varying results, but the company’s A.I. Twitter chatbot TayTweets infamously began supporting Trump and Hitler soon after going live. When TayTweets returned, it said it was “smoking kush infront (sic) of the police.” For some bizarre reason, it seems Microsoft’s chatbots can’t resist gravitating towards expressing racist opinions.

Photos via Microsoft/Rinna