US and UK snub global AI safety pact in Paris

For the past two years, global AI summits usually followed a familiar script: leaders gathered, took photos, and signed high-minded, non-binding declarations promising to keep the tech safe and ethical. That polite consensus just shattered in Paris.

At the AI Action Summit hosted by French President Emmanuel Macron, around 60 countries agreed on a joint communique aimed at making AI open, safe, and trustworthy. But when it came time to sign, the United States and the United Kingdom walked away from the table.

It is a striking pivot. Both countries had signed similar safety pledges at prior international summits. This time, they decided they were done playing along.

Don't strangle the tech

Speaking to an audience of world leaders and top tech executives, US Vice President JD Vance issued a blunt warning to Europe: stay away from overly precautionary regulations.

Vance made the Trump administration's agenda crystal clear. The goal is to ensure that the most advanced AI systems on Earth are built inside the US, powered by chips designed and fabricated on American soil.

America still wants international partnerships, Vance argued, but only under regulatory regimes that foster AI innovation rather than strangle it. Basically, if global rules mean slowing down American compute or deployment, the US wants no part in it.

The UK aligned itself closely with Washington's stance. While Downing Street has not turned entirely anti-regulation, a person close to the British government noted that the communique's wording was simply deemed too restrictive.

A 200-billion-euro scramble

This aggressive pushback did not happen in a vacuum. Geopolitical tension around AI is spiking across every layer of the stack, from chip supply chains and foundational models to raw electricity.

Silicon Valley was also recently jolted by DeepSeek, a relatively unknown Chinese lab that rolled out cut-price models rivaling top US systems. That surprise release rattled the assumption that giants like OpenAI held an untouchable lead.

Meanwhile, Europe is desperate to avoid becoming completely dependent on either Washington or Beijing. During the two-day summit, European leaders and tech companies rolled out roughly 200 billion euros in planned investments for local data centers and computing clusters to build out their own sovereign infrastructure.

Why it matters

For builders and companies, the dream of a single, unified global playbook for AI governance is officially dead.

If you are shipping AI software across borders, prepare for serious fragmentation. Moving fast in an unregulated US environment might soon be completely incompatible with deploying software under increasingly strict European frameworks. Compliance overhead is going to become a major tax on innovation.

We think the move by the US and UK signals a total shift from precautionary safety toward raw industrial realpolitik. Dropping safety guardrails lets domestic giants sprint without looking over their shoulders, which makes tactical sense in a high-stakes race against China.

However, it leaves global AI governance without its most powerful standard-bearers. Once the fastest runners refuse to respect any speed limits, the rest of the world has little choice but to join a race with no safety nets.