Geopolitics Expected to Overshadow AI Safety at Paris Summit
As political leaders frame AI as a power struggle, safety standards could take a backseat.
Tensions are high as this week’s AI Action Summit in Paris brings together heads of state, tech CEOs and researchers for the first time since the launch of DeepSeek and Donald Trump’s inauguration.
The first two international AI summits, held in the UK in 2023 and South Korea in 2024, resulted in notable (albeit non-binding) developments addressing AI safety concerns, including multinational pledges and corporate commitments.
But the backdrop of the 2025 summit is a very different tech landscape.
This week in Paris, AI safety is still on the agenda, but global politics and business interests could take center stage.
“This summit comes at a time when many are trying to position themselves in the international competition,” Macron told French newspapers on Friday. “It’s about establishing the rules of the game. AI cannot be the Wild West.”
DeepSeek’s “Wake-Up Call”
Tech stocks lost more than a trillion dollars in value last month after Chinese company DeepSeek launched an AI model developed in just two months at a fraction of the cost of US competitors.
DeepSeek’s model is free and open-source, so anyone can use and modify it. It also runs locally, meaning it can be used offline from anywhere, even without internet access.
A Peek Behind the Curtain:
DeepSeek’s spectacular breakthrough is especially notable in light of US restrictions on the sale of advanced chip technology to China.
The company’s 40-year-old founder Liang Wenfeng reportedly stockpiled Nvidia A100 chips before they were banned from export to China, and used them in combination with lower-power, unsanctioned chips to develop DeepSeek.
DeepSeek released a comprehensive paper showing how it trained its AI in a way that requires far less computing power than similar models thanks to ultra-efficient engineering optimizations.
Overinvestment?
Leading US tech giants have consistently advocated for scaling AI models through increasing numbers of chips and data centers.
During his first week in office – one week before the DeepSeek launch – Donald Trump announced $500 billion in private funding for AI infrastructure. Trump has since said that DeepSeek “should be a wake-up call” for the US tech industry.
Trump has expressed a desire to make the US the “world-capital of artificial intelligence.” Within hours of returning to the White House, he revoked Biden’s 2023 executive order implementing guardrails for AI safety, replacing it with his own AI policy designed to reduce regulatory barriers and build AI systems free of “ideological bias.”
Geopolitics in Paris
French President Emmanuel Macron and Indian Prime Minister Narendra Modi are co-hosting the two-day summit in Paris. Officials and CEOs from 80 countries are expected to be in attendance.
China’s President Xi Jinping will be represented by Vice Premier Zhang Guoqing.
Trump is sending Vice President JD Vance to lead the US delegation, marking his first trip abroad since taking office.
Notably, the US representation reportedly won’t include staff from the AI Safety Institute, which is without a director since last week.
Other attendees include European Commission President Ursula von der Leyen, German Chancellor Olaf Scholz, Canadian Prime Minister Justin Trudeau, OpenAI CEO Sam Altman, Microsoft President Brad Smith and Google CEO Sundar Pichai.
French organizers said “the summit aims at promoting an ambitious French and European AI strategy.” Since the US and China have spearheaded advancements in the sector, the summit could help make room for other players in the industry, such as French startup Mistral which also uses an open-source AI model.
The meeting could also touch on tensions surrounding EU regulation. At last month’s World Economic Forum, Trump levied “very big complaints” against the EU’s multimillion-dollar fines and antitrust penalties for US-based Big Tech companies.
The EU’s most recent AI regulation has met resistance from companies including Meta, the parent company of Facebook and Instagram.
Joel Kaplan, Meta’s Chief Global Affairs Officer, told a Brussels event last week that the EU guidelines – intended to standardize how the AI Act’s regulations are applied across the 27-nation bloc – are “unworkable” and the continent’s regulatory environment is “pushing Europe to the sidelines.”