AI Risk Debate – Findings and Market Impact
Yardeni Research reported that Anthropic identified a weapons cell operating in Yemen that employed its Claude model to develop guidance, navigation and control software for three weapons programs, including a multistage ballistic missile and a hypersonic‑glide vehicle. Anthropic stated it found no evidence that an operational weapon has been deployed.
Separately, a suspected Iran‑linked actor used Claude to analyse ship and satellite data to generate targeting intelligence on United States naval forces; the related accounts were subsequently banned and reported to authorities.
In response, Anthropic chief executive Dario Amodei and OpenAI chief executive Sam Altman have called for a slowdown in the development of more capable AI models and for the introduction of independent external evaluators operating under common safety standards.
The report notes that restrictions on closed‑source models may not apply to open‑weight systems, which can be downloaded, run privately and altered to remove safety controls. Open‑weight models from Chinese developers DeepSeek, Zhipu and MiniMax are reported to be only three to six months behind leading U.S. closed models, according to OpenRouter data, raising competitive concerns that U.S. restrictions could shift development toward harder‑to‑control systems.
The Nasdaq‑100 index fell as much as 1.5 % intraday on Monday before closing 0.5 % lower. Yardeni characterised the move as a reaction to the AI safety headlines rather than an earnings reset. Potential future signals of material impact include delayed releases of frontier models, reduced capital expenditure by hyperscalers, weaker return‑on‑investment guidance and regulators obtaining direct access to models.