The US Department of Commerce has rebranded its AI Safety Institute as the Center for AI Standards and Innovation (CAISI), redirecting its aim from general safety to addressing national security threats and avoiding “excessive and pointless regulation” internationally. Commerce Secretary Howard Lutnick announced this update on June 3rd, describing the agency’s transformation as a means to “assess and bolster US innovation” and “maintain US leadership in global AI standards.”

Originally introduced in 2023 under former President Joe Biden, the AI Safety Institute was part of a worldwide initiative to establish best practices for governments to minimize AI system risks. It formed agreements with leading US AI companies like OpenAI and Anthropic to gain early access to new models and propose enhancements prior to release. Towards the end of Biden’s presidency in early 2025, the institute issued preliminary guidelines for handling AI-related risks, which addressed the use of AI for creating biological weapons or other explicit national security threats, as well as more prevalent harmful content categories, such as child sexual abuse material (CSAM).

According to Lutnick’s announcement, the revamped institute will concentrate on tangible risks like cybersecurity, biosecurity, and chemical weapons. It will also explore “adverse foreign influence resulting from adversaries’ AI systems,” a focus area probably referencing DeepSeek, a Chinese language model that has significantly impacted the US AI sector earlier this year.