The Death of the 90-Day Review
The delicate balance between technological acceleration and artificial intelligence safety just experienced a massive paradigm shift. At the eleventh hour, US President Donald Trump completely scrapped a highly anticipated executive order on AI safety. This decision, which had been delayed multiple times, was reportedly heavily influenced by last-minute calls from tech titans including Elon Musk, Mark Zuckerberg, and former advisor David Sacks.
The now-cancelled executive order was poised to establish a voluntary review system for frontier AI models. Its most controversial clause would have mandated a 90-day window for safety reviews before any major frontier model could be released to the public. By killing the order, the administration has effectively given a green light to an unregulated arms race in the generative AI sector.
A Victory for Acceleration
The core argument pushed by Silicon Valley leaders was competitive edge. With China mapping its entire renewable energy grid using AI and aggressively pursuing Artificial General Intelligence (AGI), figures like Musk and Zuckerberg successfully argued that regulatory friction would cripple American dominance. The scrapped order was viewed by these executives not as a safety net, but as a bureaucratic anchor.
This pivot is particularly notable given the current landscape of AI development. Companies like Deepseek are actively prioritizing AGI research over short-term profits, having recently secured funding that values the Chinese startup at roughly $45 billion. In this hyper-competitive global environment, the US administration has chosen speed over caution.
Regulatory vacuums in frontier technologies do not slow down progress; they simply shift the burden of safety from governments entirely onto the shoulders of profit-driven corporations.
Why It Matters
The implications of this policy shift are profound for both the tech industry and enterprise procurement:
- Unchecked Release Cycles: Without the 90-day review period, we can expect a rapid acceleration in the release of next-generation models from OpenAI, Anthropic, and Google. This will force enterprise IT teams to adapt at an even faster pace.
- Security Vulnerabilities: As we see with advanced coding models finding zero-day exploits faster than human developers can patch them, the lack of a federal safety review could lead to powerful models being released with unforeseen capabilities or lack of proper guardrails.
- Global Fragmentation: The EU is rapidly enforcing its AI Act. With the US taking a decidedly laissez-faire approach, global enterprises will face a fractured regulatory landscape, having to build different compliance architectures for different regions.
The message from Washington is clear: the United States intends to win the AI race by running as fast as possible, leaving the safety checks for later. For developers and enterprises, this means the tools will get better faster, but the responsibility for ethical and secure deployment now rests entirely in their hands.
