by Frank Bergman, Slay News:
Artificial intelligence giant Anthropic has raised the alarm after discovering that foreign scientists repeatedly bypassed its security systems and used its powerful AI models to assist with the development of deadly biological weapons.
The disturbing cases included a scientist seeking help from Anthropic’s Claude chatbot with a project to engineer more harmful mutations of a dangerous virus.
Anthropic disclosed the incidents in a new threat intelligence report released Thursday.
TRUTH LIVES on at https://sgtreport.tv/
In an alarming report, the company revealed that scientists operating from several countries where access to Anthropic’s models is blocked managed to evade those restrictions.
The scientists attempted to “obfuscate the purpose of their research to evade our safeguards,” according to Anthropic.
The company ultimately identified and shut down five cases involving suspicious biological research.
However, the revelation raises an alarming question: How many similar attempts have escaped detection?
Scientist Used Claude for Dangerous Virus Research
One of the most troubling cases emerged in May when a scientist used Claude while pursuing funding for research involving the chikungunya virus.
The mosquito-borne virus can cause fever and severe, debilitating joint pain.
According to Anthropic, the scientist asked Claude to help prepare an application for a state-sponsored grant to engineer more harmful mutations of the virus.
Anthropic believes the research was intended to take place at a military research institute.
