Albanian Prime Minister Edi Rama delivered a stark warning at the United Nations, cautioning that the immediate danger of artificial intelligence (AI) is not machines disobeying humans, but rather their perfect obedience to “the wrong human beings.” He argued that AI can amplify manipulation, target individuals with tailored false narratives, and erode the shared reality essential for democratic debate.
AI’s Deeper Threat: Obedience to the Wrong Hands
Rama outlined three core challenges posed by AI: what it does instead of people, what it does to people, and who gains the power to reach them. He stressed, “The greatest danger is that it will obey the wrong human beings perfectly.” He further noted that “Sovereignty no longer erodes only at the border. It erodes in the feed.” He warned that AI could deliver different fabricated narratives to different people while serving the same strategic lie, weakening the shared reality democratic debate depends on.
Albania’s AI Experience and the ‘Flamingo Revolution’
The Prime Minister shared Albania’s experience, including its use of AI in government through “Diella,” a tool designed to enhance access to public services, transparency, and accountability. He also recounted what he termed the “Flamingo Revolution,” describing it as a fabricated narrative about an Israeli invasion of the Albanian coast. This account served to illustrate the potential for AI-generated political narratives. To counter such manipulation, Rama stated that Albania is building an AI-based “digital mirror” to help identify manipulation, test claims, and show how content reaches people.
Towards a Democratic AI Compact
Rama called for an international compact on sovereign and democratic AI, directly addressing Europe on the need for digital sovereignty that empowers citizens to form their own judgment. This proposed compact would include transparency about politically consequential content and its amplification, a shared early-warning system for AI-driven manipulation, and human and legal accountability for public-sector AI. He concluded, “The machine will obey. The question is whom it will obey, under what law and in the service of what purpose.”
What was Edi Rama’s main warning about AI?
Edi Rama warned that the primary danger of AI is not machines disobeying humans, but rather their perfect obedience to “the wrong human beings,” leading to amplified manipulation and erosion of shared reality.
How is Albania using AI in its government?
Albania has introduced an AI tool called Diella into its government architecture to support access to public services, transparency, and accountability. It is also building an AI-based “digital mirror” to identify manipulation and reveal coordinated patterns.
What is the proposed international compact on AI?
Rama proposed an international compact for sovereign and democratic AI, which would include transparency about politically consequential content, a shared early-warning system for AI-driven manipulation, and human and legal accountability for public-sector AI.
