The Urgency of Regulating Artificial Intelligence Against the Mirage of Private Self-Regulation

The “AI for Good” initiative, spearheaded by the International Telecommunication Union (ITU), perfectly illustrates the desire of transnational private actors to lock down decision-making spaces by blocking any attempt at binding legislation. Under the guise of digital philanthropy, the ITU is being instrumentalized as a lever for “techno-solutionism” in order to impose private industrial standards under the label of the general interest, with the sole aim of keeping States in structural dependence on the technological solutions of large capital. This body functions as a theater where the sovereignty of peoples and States is relegated to the background in favor of technicians and representatives of transnational digital corporations. These actors, omnipresent on the main stages of multilateral summits, dictate technological priorities and self-regulate without any real constraints. This deception has become glaring in the context of the defense of the rights of the Palestinian people, where the ITU, while advocating for “ethical” AI, remains silent regarding the massive use of AI systems in the surveillance of civilian populations and law enforcement activities. During this same summit, an action by pro-Palestine activists denounced the use of AI during the ongoing genocide. The Vice President of Amazon was forced to hear how the Nimbus project, which includes Lavender, allows for the tracking of individuals and the ordering of military strikes [1].

The discrepancy between the industry’s discourse, which attempts to polish the public image of AI, and the reality on the ground is striking. While developers promise technology in the service of development, AI is already being massively exploited in contexts of war, genocide, and mass crimes. The use of automated targeting systems has demonstrated, during recent conflicts in West Asia, how the automation of target designation can dilute human responsibility and transform civilian populations into “collateral damage” calculated by machines. This instrumentalization of AI for destruction does not constitute an accident, but the consequence of a technological architecture designed without safeguards, in line with an economic system that places the interests of transnational corporations above all else, socializing losses and privatizing profits. Concurrently, this technological domination is coupled with a brutal geopolitical asymmetry. The global AI market is currently an exclusive preserve, dominated by a handful of United States-based companies that hold a near-monopoly on computing power. Countries of the Global South, excluded from the levers of decision-making, are reduced to a technological dependence where their infrastructures, languages, and cultures are ignored, or even crushed by the biases of the dominant models.

We have reached a historical tipping point where AI systems cease to be mere statistical tools to become autonomous agents capable of influencing the course of armed conflicts, elections, and the global economy. Despite the magnitude of the risks, the control of this technology is currently held hostage by the communication strategies of the sector’s giants, who promote an “AI for Good” approach. Beneath this ethical veneer lies an effective tactic of normative capture: by imposing their own production standards, these companies proclaim themselves as guarantors of security, thereby thwarting any political will to legislate. This self-regulation is merely a facade intended to prevent the emergence of genuine democratic and civic control, precisely where the gravity of the stakes requires binding and universal public intervention.

It is now essential to break the cycle of unfulfilled technophilic promises. Self-regulation can no longer be considered a viable solution; it is a strategy for protecting private interests. It is urgent to initiate a multilateral, pluralistic, and democratic negotiation process, under the auspices of international public institutions, in order to establish binding regulation. It is imperative that this normative framework establish absolute transparency regarding algorithmic mechanisms (data collection, algorithms, biases, electricity consumption, copyright, etc.) and that it guarantee digital sovereignty, equal popular participation, and the inclusion of Global South States in the development of AI. Furthermore, any artificial intelligence endowed with agency (AI acting without human control) must remain subordinated to rigorous and irrevocable human supervision, while enshrining the principle that developers assume their share of legal liability for the acts generated by their systems. The management of artificial intelligence is a global common good that cannot remain the exclusive privilege of a few companies. If we do not regain control now, we risk allowing opaque management practices to take root, where machines decide, where the powerful dominate, and where international law loses all its relevance.

Faced with this deadlock, it is imperative to move beyond passivity and integrate the regulation of artificial intelligence into the draft legally binding instrument on transnational corporations (TNCs) and human rights, currently being negotiated within the UN Human Rights Council. Initiatives such as the “Pact for the Future” are limited to vague warnings with no legal bearing, promising regulations that may never materialize. The absence of an enforcement mechanism leaves the field open for companies to act with complete impunity. Compelling TNCs to answer legally for the violations and crimes facilitated by their AI systems is the only realistic way to break the cycle of impunity. It is no longer sufficient to consult experts appointed by companies; it is time to subject digital giants to a framework of criminal, administrative, and civil liability, both national and international, where their profits can no longer be dissociated from their impacts on human rights and the sovereignty of States. Without such a linkage to the binding treaty on TNCs, any discourse on AI “in the service of good” will remain an insult to the populations victimized by technological violence and a capitulation to the omnipotence of the market.

Pierre Patelli for the CETIM [1] https://www.newarab.com/news/activists-disrupt-amazon-boss-un-over-ai-services-israel