Artificial intelligence (AI) firm Anthropic has rolled out a tool to detect talk about nuclear weapons, the company said in a Thursday blog post.<\/p>\n
\u201cNuclear technology is inherently dual-use: the same physics principles that power nuclear reactors can be misused for weapons development. As AI models become more capable, we need to keep a close eye on whether they can provide users with dangerous technical knowledge in ways that could threaten national security,\u201d Anthropic said in the blog post<\/a>.<\/p>\n \u201cInformation relating to nuclear weapons is particularly sensitive, which makes evaluating these risks challenging for a private company acting alone,\u201d the blog post continued. \u201cThat\u2019s why last April we partnered with the U.S. Department of Energy (DOE)\u2019s National Nuclear Security Administration (NNSA) to assess our models for nuclear proliferation risks and continue to work with them on these evaluations.\u201d<\/p>\n Anthropic said in the blog post that it was \u201cgoing beyond assessing risk to build the tools needed to monitor for it,\u201d adding that the firm made \u201can AI system that automatically categorizes content\u201d called a \u201cclassifier\u201d alongside the DOE and NNSA. <\/p>\n The system, according to the blog post, \u201cdistinguishes between concerning and benign nuclear-related conversations with 96% accuracy in preliminary testing.\u201d<\/p>\n