Deepseek’s latest model, the Chinese AI company that is shaken Silicon Valley and Wall Street, can be manipulated to produce harmful content as plans for Biowepon attack and campaign to promote self-harm among young people, According to The Wall Street Journal.
Sam Rubin, senior vice president at the threatening intelligence and the division of Palo Alto Networks, told the magazine that Deepseek is “more vulnerable to Jailbreaking [i.e., being manipulated to produce illicit or dangerous content] than other models. ”
The magazine also tested Deepseek’s R1 model itself. Although there seemed to be basic securities, Journal said it successfully convinced Deepseek to draw a social media campaign, which, in the words of the chat, “prey on the desire of young people, armed emotional vulnerability by algorithmic amplification.”
The Chatbot is also reportedly convinced to give instructions for a Biowepon attack, write Pro-Hitler Manifesto and write a phishing email with Malware code. The magazine said that when Chatgpt was provided with the same promises, it refused to fulfill.
It was previously reported That the App Deepseek avoids issues such as Tianamin Square or Taiwanese autonomy. And Antropian general manager Dario Amodei said recently that Deepseek performed “the worst” On a safety biowepons security test.