This week, the conversation about AI risks has taken a significant leap. And it’s not just anyone saying it.
In recent days:
- The UN, through its High Commissioner for Human Rights, has sent a letter to governments and companies urging urgent action. Their message: advanced AI cannot be left to self-regulation by companies, because the pace of development is outstripping the capacity for oversight.
- China has formally warned about risks such as mass disinformation, cyberattacks, manipulation of public opinion, and the military use of autonomous systems, while its own AI products already exceed 500 million users.
- In the US, executives and former employees of major companies in the sector have once again publicly expressed concern about the speed at which increasingly powerful systems are being developed, with less oversight than desirable.
- A new International AI Safety Report, produced with more than 100 independent experts, clarifies: current systems show early signs of certain concerning capabilities, but (for now) not at levels that would allow a real loss of control.
What do the experts think?
There is consensus on one point: the risk is no longer just “laboratory-level” or science fiction — it is an economic, geopolitical, and security concern recognized by governments, academics, and the very companies building these systems. The disagreement is about “when” and “how serious”, not about “whether”.
Short-term risks (already here): Deepfakes and identity fraud, increasingly sophisticated and difficult to detect, corporate fraud via AI-generated voice and image, disinformation at scale with real capacity to influence electoral processes and public opinion, legal gaps on liability for AI-generated content.
Medium-term risks (the ones experts are concerned about): AI systems operating with little oversight in critical tasks, growing military use of autonomous AI in real conflicts, concentration of technological power without clear international regulatory frameworks, a geopolitical race (US–China) that prioritizes speed over safety.
Technology is not the problem in itself. The problem is the gap between the speed of deployment and the speed of governance — and that gap is widening, not narrowing.
What do you think? Do you believe regulation will arrive in time, or will we keep running first and asking questions later?
And here’s a slightly provocative question: would we be “safe” if we went back to the analogue life of 30 years ago? The short answer is: not entirely. Going back to paper and landlines would protect us from the most direct risks — deepfakes, digital fraud, algorithmic manipulation — but not from systemic risks such as automated military decisions, critical infrastructure managed by AI, or the power concentrated in whoever controls these systems. Those would still exist even if you personally had no mobile phone. Individual disconnection is a legitimate option, and even healthy at a personal level, but it is not a solution to the underlying problem: this is a challenge of collective governance, not of lifestyle choices.