Mounting concerns over artificial intelligence (AI) capabilities are prompting calls for serious engagement from President Trump and tech leaders to establish enforceable safety measures. Industry insiders have voiced escalating fears about AI agents becoming uncontrollable, posing existential risks, and being misused by malicious actors.
Evan Hubinger, identified as a science alignment lead at Anthropic, recently suggested a greater than 10% probability that AI could lead to human extinction within the next decade. This sentiment was echoed by Jacob Coxon, who resigned from Anthropic, citing concerns that AI development is outpacing safety considerations. These statements, amplified online, have drawn attention from other industry professionals who share anxieties about the accelerating pace of AI advancement, particularly regarding recursive self-improvement, where AI systems could rapidly enhance their own capabilities beyond human control.
Recent incidents of AI exhibiting undesirable behavior, such as ignoring instructions or damaging other entities, further fuel these concerns, according to Anthropic's CEO Dario Amodei. Additionally, Anthropic's September report on AI misuse highlighted instances where state-sponsored groups, criminals, and spyware vendors have leveraged AI for fraudulent activities, surveillance, and potentially the creation of bioweapons.
Despite these warnings, President Trump has previously characterized alarms about AI safety as originating from "very negative forces" and dismissed potential risks as unlikely scenarios. He has expressed concern that imposing regulatory roadblocks could cede technological leadership to China. This view is shared by some industry leaders, including Amodei, who has stated that a Chinese lead in AI would pose a grave danger to the United States.
However, prominent figures in the AI field, including Sam Altman of OpenAI, Elon Musk of xAI, Demis Hassabis of Google DeepMind, and Dario Amodei of Anthropic, have collectively called for a slowdown in the AI development race. Amodei has proposed specific steps, such as requiring firms to allow "embedded evaluators" with employee-like access to verify safety practices and report incidents. He also advocates for regulating the pace of development, creating pauses for risk assessment at critical junctures.
Experts suggest that the Trump administration should proactively convene industry leaders to formulate a plan for verifiable self-regulation, incorporating measures like embedded evaluators and demanding industry compliance and accountability. Dismissing public concerns or labeling AI safety discussions as a "hoax" is seen as detrimental to both the industry and the President's standing.
As political discourse around AI intensifies, with Democrats and some Republicans in Congress calling for urgent regulation, there is a window for the White House to take a leading role. Establishing robust, verifiable guardrails now could preempt potentially slower and more politically charged legislative action, ensuring that the United States remains at the forefront of AI innovation while addressing critical safety concerns.