There is a tendency in conversations about AI to treat it as primarily a technical challenge — a problem that engineers will solve if given sufficient time and compute. Emmanuel Macron’s speech at the AI Impact Summit in Delhi was a corrective to this tendency. The French president argued, with characteristic directness, that AI is also a political and moral challenge, and that the people best equipped to address that dimension are not engineers or billionaires but democratically elected governments.
His evidence was hard to dismiss. Research by Unicef and Interpol found that in a single year, 1.2 million children in 11 countries had been victimised by AI-generated explicit deepfakes. These are crimes made possible by technology that is legal, available and rapidly improving. The companies that build and deploy this technology have not solved this problem, despite years of stated commitment to safety. Macron’s argument is that they cannot solve it alone — that only governments with the power to legislate and enforce can create the conditions for genuine accountability.
Macron was also pushing back against a specific political argument from the United States. The Trump administration has framed AI regulation as a threat to American competitiveness and entrepreneurial freedom. Macron’s reply is that this framing presents a false choice: Europe regulates AI and continues to innovate and invest. Safety and dynamism are not opposites. What Europe refuses to do is treat the harm caused by unregulated technology as an acceptable price of progress.
The summit offered several contrasting visions of AI’s future. António Guterres called for a global commons approach. Narendra Modi advocated open-source development. Sam Altman painted a picture of AI systems that would soon exceed human intellectual capacity — a prospect that makes the current governance deficit even more alarming. Dario Amodei raised concerns about autonomous AI behaviour and its potential for misuse. Each of these visions implies different governance arrangements, but all of them imply some form of governance.
Macron’s contribution in Delhi was to insist that child safety is not a narrow issue to be resolved by content moderation teams inside tech companies. It is a test case for the entire project of democratic governance in the AI era. If governments cannot protect children from documented, large-scale harm, what exactly is their authority for? That question does not have a comfortable answer, which is precisely why Macron keeps asking it.
Macron’s Moral Clarity in Delhi: Why AI Needs Governments, Not Just Engineers
