[Anthropic Discloses Dangerous Use Cases of Claude, Claims Iranian Accounts Threatened U.S. Military Targets] According to a report by Golden Finance on September 11, U.S. AI company Anthropic stated on Thursday that it had prevented certain scientists from using its Claude AI model for "potentially supporting biological weapons development." This information was revealed in a detailed report. The report listed five cases, one of which involved research on how to enhance the transmission capability of a mosquito-borne disease. Anthropic noted that it is difficult to definitively determine whether these cases were related to plans for biological weapons development but emphasized that the company had blocked the related activities and reported the information to relevant authorities and industry partners. The report also disclosed other behaviors involving surveillance, fraud, conventional weapons development, and propaganda activities. Anthropic claimed to have identified and banned accounts using Claude to conduct "influence operations" aimed at shaping public opinion, including three accounts linked to Iran. Anthropic revealed that one of the Iranian-linked threat actors used Claude to formulate attack target recommendations against U.S. Navy forces in the region.