US Drops ‘Safety’ From AI Strategy to Secure Global AI Leadership
Why did the US remove "safety" from the AI Safety Institute? What is the new direction for America's AI policy? If you're searching for answers about the United States’ evolving stance on artificial intelligence governance, the short answer is this: the AI Safety Institute has been rebranded as the Center for AI Standards and Innovation (CAISI), shifting its focus from general AI safety to national security, AI competitiveness, and tech leadership on the global stage. This dramatic policy pivot underscores a growing effort by the US to dominate AI standards, while minimizing regulatory hurdles and maximizing economic growth from AI technologies.
The Department of Commerce announced this major overhaul on June 3, 2025. According to Commerce Secretary Howard Lutnick, CAISI will now center its mission around protecting the country from AI-related threats such as cybersecurity breaches, bioweapons, and foreign AI influence, rather than focusing on broader ethical or societal risks. This reflects a wider shift under the current administration to prioritize AI innovation, defense, and international standard-setting over prior safety-focused measures.
This transition moves away from the AI Safety Institute’s original goals set under former President Joe Biden. Launched in 2023, the Institute had forged agreements with leading AI companies like OpenAI and Anthropic, gaining early access to AI models to evaluate risks. Its guidelines had addressed issues like the use of AI to produce child sexual abuse material (CSAM) or assist in biological weapons development. Now, under new leadership, CAISI is narrowing its lens to only “demonstrable risks” — mainly national security threats — while shedding the broader and more controversial realm of AI ethics regulation.
Critically, the announcement aligns with President Trump’s broader strategy to accelerate American dominance in the AI sector. His administration has repealed several Biden-era executive orders that focused on AI safety and regulatory oversight. Instead, it has launched initiatives encouraging AI adoption in education, energy-hungry AI data centers powered by coal, and stripped back regulation across states — a move designed to fuel economic development and private sector growth in AI.
Among the high-impact targets of CAISI’s new mission is DeepSeek, a powerful Chinese large language model that made headlines earlier this year by outperforming many Western counterparts. The agency will now investigate foreign AI systems deemed hostile or manipulative, aiming to safeguard US infrastructure and influence in the increasingly competitive global AI race.
This transformation raises pressing questions: Will minimizing safety oversight in favor of dominance backfire? Or is this a necessary evolution for the US to stay ahead in the AI arms race? As global AI innovation accelerates, the answer may shape the future of tech regulation, cybersecurity, digital policy, and international diplomacy.
Comment