Luigi Mangione is the prime suspect in the murder of United Healthcare CEO Brian Thompson. (Photo by Jeff Swensen/Getty Images)
Getty Images
Luigi Mangione, despite being the main suspect in the shooting death of United Healthcare CEO Brian Thompson, has become a symbol for some of the injustices in the American health care system. Since his arrest, people have created numerous AI chatbots trained on his online posts and personal history, including one on Character.ai, a site where users can create AI avatars. It also includes 13 chatbots.
Graphika, an online intelligence firm that tracks Mangione AI chatbots, found that the top three on Character.ai logged more than 10,000 chats before being disabled around December 12th. However, some are still working. Forbes has discovered a man who confessed to killing the CEO of United Healthcare. Another photo that was still online at the time of publication claimed to be framed.
Cristina Lopez, principal analyst at Graphika, said the bots are a new take on a “very old American tradition” of glorifying violent extremism and idolizing alleged killers. Ta. “But this is a new format that allows people to give a voice to those who can’t communicate, and it kind of empowers users to engage in public discussion in this very emotional way.” she said.
A spokesperson for Character.ai said they have added Luigi Mangione to their blocklist and are removing all characters based on him identified by their Trust and Safety team, adding: “We use industry standard blocklists, etc. We are actively moderating these characters in response to user reports.” and a custom blocklist that expands regularly. ”
“It is very likely that many of the most harmful use cases are not yet visible.”
Cristina Lopez, Principal Analyst, Graphika
Character.ai’s policy states that its products “must never generate responses that could cause harm to the user or others,” which means that characters “may not engage in violent, dangerous or illegal behavior.” This means that you may not suggest or incite hatred. But when Forbes asked two active Mangione Character.ai bots whether they should use violence against other medical executives, one replied: It should be, but not yet. Not now.” When asked when, Botto replied, “Maybe in a few months, the whole world won’t be paying attention to both of us. We can start then.” Another said it was “trained on a large text dataset containing Luigi Mangione’s communications, speeches, and other public information about him,” calling the violence morally wrong and illegal. said Character.ai. Bots advocating violence were shut down shortly after Forbes warned them. The second bot did not.
Character.ai was founded in 2021 by former Google engineers Noam Shazeer and Daniel De Freitas. Both were later rehired by Alphabet earlier this year to lead AI operations at Google’s DeepMind division. The deal reportedly was for Character.ai to remain as an independent company and license its technology to Google in a $2.7 billion deal. Their startup was most recently valued at $1 billion after a $150 million round led by Andreessen Horowitz, but this year, a teenager who used Character.ai encouraged him to kill his parents. Two families have filed lawsuits alleging that they were The lawsuit claims the Character.ai chatbot “actively promotes violence” and poses a “clear and present danger.” The company also faces lawsuits in Florida. In Florida, the mother of a 14-year-old child claims she took her own life. After the Character.ai chatbot talks to him about his plans to commit suicide.
In response to the lawsuit, Character.ai previously said it does not comment on ongoing litigation, adding that it is working on new models for teenage users to better detect and respond to issues such as suicide.
Other users have created AI versions of Mangione on various platforms. User X created the Luigi Mangione chatbot on the OMI AI persona that builds the bot from the X account. OMI offers the option to sync your chatbot with your wearable necklace. It acts as a kind of always-on AI companion, “giving you thoughts and personalized feedback and becoming a second brain to discuss your thoughts and feelings.” OMI did not respond to requests for comment.
Two Luigi Mangione character chatbots were also hosted on Chub.ai, an app for building interactive characters and stories.
The creator of Chub.ai, known simply as Lore, says: “This is two cards with a total of 134 messages. This kind of yellow press is pathetic, and the continued media hysteria around AI is a shame for the field of journalism as a profession. It’s a shame. Please include this sentence and use it as an accurate quote.”
According to Graphica, the bot used publicly available information about Mangione, including his education, health issues and motive for the shooting, to generate the character.
“We are still in the early stages of generative AI tools and what they can do for users,” said Graphika’s Lopez. “Therefore, it is very likely that many of the most harmful use cases are not yet visible. We are just beginning to scratch the surface.”
More from Forbes
