When Truth Becomes Negotiable: AI’s Impact on Epistemology

4 min readOct 1, 2025

Picture this: a courtroom in 2030 where both the prosecution and defense bring videos as evidence, each indistinguishable from reality, each showing the opposite story. Which one does the jury believe?

That’s what’s on the horizon. As a pseudo-demonstration of this, OpenAI’s Sora 2 release doesn’t just blur the line between real and fake. It makes truth itself negotiable.

OpenAI’s Sora 2 demo video

Many people will celebrate the cinematic brilliance, world-class sound, and the fact that small creators now have Hollywood-scale tools. Others will sound the alarm over job loss, deepfakes, and the terrifying ease of manufactured realities. But here’s the twist: we’re not just talking art, jobs, or reality; we’re talking a seismic shift in the currency of belief.

The deeper disruption is in Epistemology, which is the investigation of what distinguishes justified belief from opinion. And Sora 2 drops us right into that contested arena. When any reality can be spun up on demand, the question shifts from what is true to who gets to decide what counts as truth. We move from debating facts to debating authority. The scarce resource is no longer creativity or compute power, but credibility itself.

Once machines can conjure visuals, audio, interaction, memory — all indistinguishable from what is “real” — truth stops being a fixed reference. It becomes a contested artifact.

The Vanishing Anchors of Trust

That shift in power is not subtle. For most of human history, trust has been anchored in shared references: eyewitnesses, institutions, publications, lived memory. Even when those anchors were imperfect, they gave society a common baseline for ‘what happened.’

Now imagine a world where the anchors themselves can be fabricated with precision. A video of an event, a voice message, or even a photograph; none can be taken as proof without verification. Which means the authority doesn’t lie in the content but in the credential.

Whoever controls the stamp of authenticity (whether it’s a platform, a government, a corporation, or an algorithm) effectively controls the story. They decide which version becomes the ‘official’ memory, which narrative earns trust, which identity is preserved or erased.

And history gives us clues that this has happened before.

Roman generals minted coins and built monuments to enshrine their victories while erasing defeats.
Medieval monarchs issued decrees through scribes, ensuring their narrative outlasted dissent.
World War II propaganda ministries rewrote public memory with posters and films that made the unbelievable believable.
Corporations in the 20th century defined consumer identity with advertising campaigns that blurred desire into necessity.

Back then, controlling truth required armies, presses, or entire industries. Today, it can be done with a model prompt. The implications for who we trust, and why we trust them, are staggering.

A Historical Pattern of Power

Seen this way, SORA 2 is not an outlier but the latest in a long chain of truth shifts. Each era has had its referees of reality:

Scarcity Era: When manuscripts were copied by hand, the Church decided which texts survived and shaped a new collective memory.

Access Era: The printing press democratized narrative but also unleashed disinformation, which included religious wars and centuries of contested authority.

Reach Era: Broadcast media amplified voices at scale, but those voices were tightly controlled by governments and corporations.

Synthesis Era (Today): AI no longer merely distributes information; it can fabricate it. Truth is not scarce, nor is it limited by access or reach. It can be infinitely manufacturable.

The danger is not only deception. It is the consolidation of power in whoever owns the verification layers.

The Coming Economy of Authenticity

Where there is scarcity, there is an economy. The new scarcity is credibility.

Platforms will monetize authenticity. We already see this in social media verification badges, but the next wave could be AI-driven authenticity seals embedded into content. What this leads to could be a ‘pay to prove you’re real’ model, not too different from the pay-to-play ‘verified’ user checkmarks on X (formerly Twitter).

I believe that corporations will shift to build entire markets around ‘authenticity as a service.’ Your videos, contracts, or memories might require third-party validation before being accepted.

Governments may legislate ‘truth authorities,’ perhaps even mandating which verification protocols count in court or commerce. (This has happened before in analog ways: “let me see your papers to pass this checkpoint”).

This economy may be lucrative, but it will also be fragile. Because once trust is monetized, trust itself becomes transactional.

The Human Layer

This isn’t just systemic, either; I believe it will inevitably touch personal lives.

How will families know if their heirloom photos or videos of grandparents are authentic or synthetic?
How will courts handle “evidence” when nearly anything can be manufactured with precision?
What happens when an entire community is gaslit by fabricated events they never lived through?

Memory is identity. If our memories can be rewritten with the same fidelity as fact, then identity itself becomes vulnerable. We may find ourselves asking not just, “What happened?” but “Am I who I remember being?”

So what do we do in an age when truth is negotiable?

We Need New Norms

First, it seems to me that we will need new literacies. Just as society once taught reading, writing, and media literacy, we now need epistemic literacy: the ability to interrogate not just the message but the source of credibility. (And I think we’re here right now).

Second, we will need new institutions. Fact-checkers and journalists were built for the broadcast era. In the synthesis era, we may need distributed verification networks, where authenticity is confirmed through consensus and cryptographic proof, not centralized gatekeepers. And even this is problematic, because who checks the gatekeepers?

Third, we will need new norms. Social trust may no longer be about assuming content is true until proven false, but the reverse: assuming it is synthetic until proven authentic. That cultural inversion will feel jarring, but it may become essential.