Artificial intelligence could significantly simplify the development of weapons of mass destruction, according to Dmitry Stefanovich, a researcher at the Center for International Security at IMEMO RAS.

Stefanovich said the issue is particularly serious in relation to nuclear weapons and other forms of WMD, where AI-based systems could noticeably lower the complexity of certain development tasks.

He noted that reports from the U.S. company Anthropic already include recurring examples of both state and non-state actors using artificial intelligence to develop or refine various weapons systems, including missiles and drones.

At the same time, Stefanovich stressed that AI cannot replace a full scientific and engineering base or real-world testing. However, he said the technology can accelerate and optimize individual stages of the development process.

Earlier, Anthropic researcher Jacob Coxon announced his departure from the AI industry, citing concerns that competition between technology companies was pushing the world toward a «point of no return» and could lead to an apocalyptic scenario by the end of 2027.

Anthropic CEO Dario Amodei later called for slowing the development of more powerful AI models, pointing to safety risks.