#7 Society and Technology Bulletin

class=\

The human voice has begun to lose its unique quality that no other living being could replicate. Our voices are now nearly indistinguishable from real ones, capable of being imitated with high fidelity. A recent study by Consumer Reports reveals that leading artificial intelligence models offer almost no protection against the cloning of human voices without the individuals’ knowledge or consent.

In daily life we use our voices with little restraint. As speech is one of our most fundamental abilities, it has become encoded as the representative of all our public expressions. In the modern era, our voices are recorded and processed in countless ways. For example, calls with customer service are recorded “to improve service quality.” Digital assistants are increasingly exposed to our voices. When we post voice or video content on social media, we rarely consider the broader implications.

Yet our voice is far more than a communication tool—it is an inseparable part of our personal identity. We externalize many aspects of our existence—our actions, commands, ideas, and suggestions—at the level of our voice. The use of our voice is thus a form of existential expression.

In recent years, voice cloning technology has become possible thanks to AI-powered systems that can replicate a person’s voice using just a few seconds of audio. Once considered science fiction, this technology is now used across many domains, from marketing to political propaganda. Last year, a fake audio clip of Joe Biden was used in the U.S. Democratic primary to discourage voters from casting their ballots, exposing the potential for abuse of this technology.

Today, in many places, no special permit or certification is required to use voice cloning technology. With open-source models or commercial applications, anyone can take a social media video, extract the person’s voice within minutes, and generate synthetic speech to make it say anything desired. There is almost no legal framework to distinguish between parody, art, and fraud. Thus, voice security is left entirely to the goodwill of companies and the prudence of users.

This situation has triggered a serious crisis of privacy and authenticity. Cloning someone’s voice, speaking on their behalf, representing them, and instrumentalizing their influence can all be seen as violations. At this point, the fundamental question we must ask is whether privacy is merely a matter of physical distance. With the capabilities offered by artificial intelligence, every environment where we share our voice becomes a ground upon which we surrender our privacy.

A common reflex in the tech world to address privacy crises is to solve the problem with the very technology that caused it. Emerging systems aim to embed digital authenticity markers (watermarks) into voice files to determine whether a recording is original or synthetic. Like security features used to detect counterfeit currency, these invisible signatures are being developed to identify AI-generated voices.

A more advanced proposal involves linking blockchain technology to voice identities. In this system, a unique “fingerprint” of a person’s voice is recorded on a distributed ledger, and every voice generation is verified against this record. This approach opens up a new domain of digital identity: cryptographically securing, publicly documenting, and making immutable the origin of a voice.

The appeal of this approach lies in its promise to rebuild trust. But we cannot avoid asking: Should the solution to privacy lie in greater traceability? Are we legitimizing more surveillance in the name of security? Does proving that a voice truly belongs to its owner impose a new order that restricts the voice’s inherent freedom to exist? This path constructs the infrastructure of a world where humans are defined as data-owning entities. The more we entrust security to technology, the more we tie trust not to intent but to traceability.

Another reflex is to believe nothing.

As in the cybersecurity principle of “Zero Trust,” no user, device, or access request should be trusted by default; continuous verification is required. If we constantly keep in mind the possibility that a voice may be synthetic, this may make us not only more cautious but also more suspicious, more isolated, and more emotionally detached. If we approach every voice with caution, do we undermine the minimal trust that makes social life possible?

Do we want to live in such a world? In a society where every video is fake, every voice is artificial, and every piece of information is manipulable, how possible are public communication, emotional connection, and authentic debate?

Alongside the rise of artificial intelligence, digital data centers have transformed into energy giants comparable to the iron and steel factories of a bygone era. According to the International Energy Agency’s (IEA) 2025 report, the electricity demand of data centers alone will quadruple by 2030. Projections suggest this demand will exceed Japan’s annual energy consumption.

This can be interpreted as the “invisible body” of artificial intelligence. Every computation performed for large language models, image processing systems, or autonomous vehicle algorithms requires substantial energy on server racks. Behind every “virtual” command lies a physical energy cost.

While AI competition has long been defined by algorithmic superiority, data volume, and model capacity, this framework is now inadequate. High-performance AI applications such as large language models, image processing systems, and autonomous decision networks demand not only cognitive resources but also physical ones.

An increasingly decisive factor: energy. Progress in artificial intelligence is becoming increasingly dependent on energy infrastructure. This transforms the AI race from a purely technical competition into a comprehensive energy rivalry with geopolitical and ecological dimensions.

In the AI competition between the U.S., Europe, and China, energy sources that power these systems are becoming strategic assets. It is known that the U.S. must expand its electrical grid by 20 percent to sustain its data center infrastructure. The Stargate project, a partnership between OpenAI, Oracle, and SoftBank, signals this kind of growth. Meanwhile, France is positioning its nuclear energy infrastructure as the cornerstone of its AI strategy.

As in the past, energy continues to play a strategic and decisive role in global power competition. We will see whether an energy supply crisis brings about a classic geopolitical reckoning.

Arms control experts, activists, and some representatives of major tech companies argue for an “ethical brake” on the breakneck pace of the AI race. At the AI Security and Ethics Conference hosted by UNIDIR (United Nations Institute for Disarmament Research) in Geneva, this call became even clearer: Artificial intelligence is approaching its Oppenheimer moment.

Robert Oppenheimer’s role in developing the atomic bomb and his subsequent remorse represent a historical turning point demonstrating how technological genius, when unbounded by moral limits, can lead to catastrophic destruction. The “Oppenheimer moment” analogy for AI suggests we are now at a similar threshold: a point where the gap between what is technologically possible and what is ethically permissible has opened wide.

UNIDIR officials emphasize the need to strictly regulate AI-driven military technologies, particularly autonomous weapons systems, to ensure they comply with human rights, international law, and the ethics of war, as unchecked AI can make inhuman decisions.

Europe’s stance in the AI race offers a striking example in this context. Influenced by Europe’s intellectual heritage, the tradition of “normative thinking” sometimes seeks to set direction even at the cost of slowing technological momentum.

This creates serious decision moments for Europe. The dilemma is particularly evident regarding AI’s role in military applications: on one hand, moral risks of autonomous weapons are condemned; on the other, rapid advances by actors such as the U.S., China, and India create a risk that Europe will fall behind in terms of national security and strategic capacity.

A decision paradox emerges here:

• Either slow technological progress due to ethical concerns,
• Or set aside ethical concerns and enter the race for military superiority.

This is not merely a state-level dilemma but an epistemological quandary faced by individuals as well.