Beyond AI Detection: Toward a New Model of Truth Verification

Today's debate about artificial intelligence tends to focus on detecting AI-generated content. That approach, however, is limited and, in the long run, ineffective. The real problem is not whether a piece of content was created by an AI or by a human, but whether that content is false, manipulative, or socially harmful. This article proposes a paradigm shift: moving from AI detection to validating authenticity, relying on verification and traceability systems and, potentially, on a strategic return to analog processes certified by hardware.

1. Introduction

The rapid progress of generative models has created growing concern about the proliferation of synthetic images, videos, text, and audio. The dominant response has been to build tools that detect AI-generated content. This strategy, however, starts from a questionable premise: that the artificial origin of content is, in itself, the problem.

This piece argues that the premise is wrong. The origin of content is secondary to its impact and its truthfulness.

2. The false dilemma: AI vs. human content

Not all AI-generated content is problematic. There are creative, artistic, and personal uses that pose no social risk at all. Likewise, a large share of false, manipulative, or malicious content has historically been produced by humans.

We are entering the era of the "Liar's Dividend." When anything can be fake, anyone accused of something real (a video of a crime, for example) can simply claim it was created by AI. The lack of verification systems does not only let the false pass as true; it lets the true be dismissed as false. This erodes the foundations of the justice system and of public debate.

Centering the debate on whether something was created by an AI therefore introduces a false dilemma. The relevant question is not "who or what created it?", but:

• Is it true?
• Is it verifiable?
• Does it have social, political, institutional, or legal consequences?

3. Why detection is bound to fail

In the short term, humans and algorithms can identify certain patterns in AI-generated content. That advantage, however, is temporary. As models improve, detection becomes increasingly imprecise.

It is reasonable to anticipate a scenario in which:

• Human detection becomes unreliable.
• Automatic detectors produce more false positives than correct results.
• Synthetic content becomes indistinguishable from human content.

Basing social trust on a system destined to fail is not a sustainable solution.

4. Changing the focus: from detection to verification

The alternative is to shift the focus toward validating content. Instead of trying to identify what is fake after it has been published, we must be able to prove what is authentic from its origin.

This is especially critical in contexts such as:

• Evidence of crimes.
• Journalism and historical documentation.
• Scientific research.
• Institutional statements.
• Testimony and legal records.

The authenticity of content is no longer just a matter of aesthetics or copyright; it is a matter of national security. The ability to fabricate official statements or evidence of armed conflicts requires states and institutions to adopt "Zero Trust" protocols, in which no information is processed unless it comes with a certificate of origin based on certified hardware.

In these cases, trust cannot depend on how content looks, but on its traceability.

Interactive // Detect or verify?

Three scenarios. Choose how you would respond and compare your decision with the article's argument.

Scenario 1. A real video of a crime is circulating. The accused person claims it was created by AI.

Run it through an AI detector Check its certificate of origin and chain of custody

Scenario 2. An AI-generated illustration for a personal art project.

It's a problem: an AI made it No risk: the question is whether it's true, verifiable, and consequential

Scenario 3. An institution receives an alleged official statement from another state, with no certificate of origin.

Process it if it looks authentic Don't process it until it comes with a certificate of origin

5. The role of hardware and the return to analog

One possible solution is to reintroduce analog or hybrid processes as authentication mechanisms. For example:

• Cameras with physical film and verifiable development processes.
• Typewriters or devices with no connectivity.
• Dedicated hardware that cryptographically signs content at the moment it is created.

Unlike bits, which can be cloned infinitely, chemical and mechanical processes have "unique imperfections." The grain of a specific photographic film, the imprint a typewriter's typefaces leave on paper, or the micro-variations in the speed of an audio tape act as forensic DNA. In a world saturated with synthetic perfection, the asymmetry of physical imperfection becomes the gold standard of proof.

The goal is not technological nostalgia, but structural trust. Hardware can act as a witness to the creative process, something software alone cannot guarantee.

6. Blockchain and traceability systems

Once content has been created in a trusted environment, technologies such as blockchain can be used to:

• Record the moment of creation.
• Guarantee the integrity of the content.
• Document the chain of custody.
• Provide transparency and public auditability.

The future is not purely analog, but physical-digital. Imagine cameras with "secure enclaves" (chips similar to the ones that protect banking data) that generate a cryptographic hash at the exact moment light hits the sensor (or the physical shutter is triggered). That hash would be uploaded to a decentralized network (a blockchain), creating an unbreakable link between the physical event and the digital record. Hardware thus becomes a "technological notary."

Diagram // Hardware as a technological notary

1. 01Physical eventlight hits the sensor
2. 02Secure enclavegenerates a cryptographic hash
3. 03Decentralized networkrecords the moment and integrity
4. 04Verificationan auditable chain of custody

This does not validate the "truth" of the content, but it does establish a solid framework of authenticity and accountability.

7. Conclusion

Artificial intelligence does not destroy truth; it destroys the traditional methods for certifying it. Insisting on detecting AI is a reactive, short-term response. The real challenge is to rebuild systems of trust suited to an era in which synthetic generation is trivial.

We must stop chasing the shadow (AI) and start strengthening the object (authenticity). The paradox of our era is that, to save our digital civilization, we will have to recover and shield our physical certainties. Technology will not save us from technology; only a new ethics of verification, backed by physics and cryptography, can do that.

The future will not depend on identifying what is artificial, but on being able to prove what is authentic. The problem is not AI. The problem is how we verify reality in a world where seeing it is no longer enough.