Security experts Melanie Sisson and Tianjiao Jiang have proposed nuclear-style safeguards for military AI, urging the US and China to bar autonomous systems from launching cyberattacks on nuclear command networks or initiating nuclear launches. The unofficial proposals, emerging from a long-running Brookings-Tsinghua dialogue, call for a dedicated military hotline and a shared definition of human control. The recommendations arrive amid a broader debate over slowing frontier AI development.
US-China military AI safeguards proposal
- ▪Neither the United States government nor the Chinese government has endorsed the military artificial intelligence safeguards proposed by Melanie Sisson and Tianjiao Jiang.
- ▪Security experts Melanie Sisson of the Brookings Institution and Tianjiao Jiang of Fudan University proposed a set of military artificial intelligence safeguards modeled on nuclear weapons rules.
- ▪The nuclear-style military artificial intelligence safeguards proposal authored by Melanie Sisson and Tianjiao Jiang emerged from the US-China AI and National Security Dialogue, a track run since 2019 by the Brookings Institution and Tsinghua University
Nuclear command system protections
- ▪Failures in nuclear command, control, and communications systems could disrupt communications, misread attacks, encourage first strikes, or trigger accidental or unauthorized launches.
- ▪The nuclear-style safeguards for military AI proposed by Melanie Sisson and Tianjiao Jiang would bar artificial intelligence from autonomously deciding to use nuclear weapons or attacking a rival's nuclear command, control, and communications systems
Human control requirements
- ▪The nuclear-style safeguards for military AI proposed by Melanie Sisson and Tianjiao Jiang build on a November 2024 agreement where Joe Biden and Xi Jinping affirmed that humans must retain control over nuclear weapons decisions
- ▪The nuclear-style safeguards for military AI proposed by Melanie Sisson and Tianjiao Jiang require that humans remain in control of any consequential cyberattack targeting a rival's nuclear infrastructure or critical infrastructure
- ▪A proposal by security experts Melanie Sisson and Tianjiao Jiang urges the United States and China to agree on a shared definition of "meaningful human control" over weapons systems
Military AI incident hotline
- ▪Carla Freeman of Johns Hopkins University stated that a proposed dedicated US-China military hotline for AI incidents may be delayed because Chinese officials avoid communicating with the United States until top leadership settles on a response to an incident
- ▪Security experts Melanie Sisson and Tianjiao Jiang proposed establishing a dedicated United States-China military hotline to manage incidents involving autonomous artificial intelligence
AI slowdown debate
- ▪China's commerce ministry dismissed proposed artificial intelligence development slowdown plans as a bid by the United States for technological hegemony.
- ▪Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and xAI owner Elon Musk have supported calls for developers to slow down frontier artificial intelligence capability gains.
Story comments
Loading comments…