Dario Amodei, chief executive officer of Anthropic, recently published an essay urging companies in the artificial intelligence sector to slow the pace of capability development in response to growing concerns about AI risks. Former US President Donald Trump responded to those remarks on his social platform, Truth Social.
In his post, Trump argued that the only «control mechanism or ‘safety constraint’» AI needs is a strong and smart (high‑IQ) president, and asserted that the United States already has such leadership. He claimed the Trump administration prevented AI «actors» from doing harmful or potentially harmful things and added that his administration already wields substantial criminal and regulatory power over companies.
Trump also warned of what he called a "sick conspiracy" against artificial intelligence and data centers that, he said, benefits China. He wrote, "WHOEVER WINS Artificial Intelligence, WINS! We lead China and everyone else, and that will continue into the future." He concluded by threatening those he described as conspiracy believers, traitors and leakers.
Anthropic’s allegations and researcher concerns
Anthropic has reported that multiple market actors misused its Claude language model, alleging that Claude was used for weapons development, surveillance and fraud. The company specifically accused Moonshot, a major Chinese AI firm, of secretly forwarding incoming requests to Anthropic’s chatbot Claude and using Claude’s responses to train its own language model. Anthropic said Deepseek and Xiaomi engaged in similar conduct.
The dispute was further intensified when Jacob Coxon, an Anthropic researcher, resigned, stating that "those building artificial intelligence seriously believe it could kill us all by the end of the decade." Coxon’s departure and Anthropic’s allegations have spurred debate over AI safety, whether development should be slowed, and the need for regulatory action.
Why this matters
The episode matters both technologically and politically. Technologically and ethically, it raises questions about how quickly capabilities in large language and general‑purpose models should be increased. Politically, it illustrates how senior politicians—including Donald Trump—are framing AI governance in terms of personal leadership and national competition. Anthropic’s claims, the researcher’s resignation, and the accompanying geopolitical rhetoric underscore that AI safety and industry accountability are issues with implications well beyond the tech sector.
