Zelensky’s “$200 million suit scandal”: When oracles become tools for making money, how does AI solve the “truth dilemma”?

By Omer Goldberg

Compiled by: Tim, PANews

The controversy over Zelensky’s suit on Polymarket was not a glitch. It was a $200 million case that exposed a fundamental flaw in human-controlled oracles: when the cost of corruption is lower than the reward, facts become commodities purchased by the highest bidder.

Zelensky's $200 million fashion show

Picture this: Zelensky walks into a NATO summit wearing what all major media outlets call a suit. With $200 million in volume on the prediction market, the results seem obvious.

But the UMA oracle gave a "no" in the prediction of "whether Zelensky will wear a suit in July"

It's not because he wasn't wearing a suit, nor because the evidence was insufficient.

Because those who control the oracle have bet tens of millions on the “no” option, they can rewrite reality simply by using their voting power with almost no real risk.

Oracle manipulation

The uncomfortable truth about human-controlled oracles is that humans are biased.

• Some top UMA token holders are betting heavily on “no”.
• When “yes” looked like the correct outcome, instead of accepting the loss, they began to manipulate the vote.
• Over 23 million UMA tokens (worth approximately $25 million) were staked to counter the outcome.

This is not decentralization, this is purely whales protecting their positions.

As long as there are enough UMA tokens and operations, the facts no longer matter, only the results.

The Wider Oracle Crisis

This problem has implications far beyond Polymarket and UMA. Oracle systems controlled by humans are susceptible to a variety of manipulation methods and face various pitfalls and flaws in incentive mechanism design.

Although we use the Zelensky suit incident as a case study, it is important to point out that we have previously observed this problem in the case of the Ukrainian mineral transaction in March 2025.

All major prediction markets face the same fundamental challenges.

When humans control the right to define the truth, the truth becomes a tool for profit-making.

The evolution of oracles: from human control to intelligent decision-making

The only real solution to the problem of human-controlled oracles is to remove human subjectivity entirely.

AI-driven oracles will change this:

• No financial incentives: The model neither holds positions nor cares about who gets the final outcome.
• Anti-bias decision rules: Same training weights, prompt words and temperature parameters = the model will score the evidence based on the same underlying criteria. AI has no emotional fluctuations, no off-site interests, and no behind-the-scenes transactions.
• Reasoning pipeline: Each intermediate process step can be recorded, reviewed and replayed.
• Machine-level throughput: Thousands of data sources can be processed in parallel without resting or relying on any human intervention.

Residual error still exists, but it is random statistical noise. This error is extremely difficult for traders to exploit. With clear solution standards and certified data sources, the current state-of-the-art models have production-level accuracy, and the accuracy curve is showing a steep upward trend.

Residual noise is better than a calculated lie

The future of prediction markets must completely exclude humans from determining the truth.

The specific form of this architecture is as follows:

• Predefined source hierarchy: Reuters > BBC > Local News > Blogs
• Cryptographic proof of data origin: ensuring that the information has not been tampered with
• Multi-agent consensus: multiple AI systems each reach independent conclusions
• Traceable reasoning: Every decision has a complete audit trail
• Immutable evidence: Proofs stored on the blockchain cannot be modified or deleted

Determining the truth in the post-truth era

Prediction markets are a microcosm of a larger challenge. When Wikipedia can be edited, news can be tampered with, and “facts” become negotiable, we need to build systems that can establish objective truth.

This problem has implications far beyond prediction markets themselves:

• Election Integrity and Authentication
• Scientific consensus and research verification
• News authenticity verification in the era of deep fakes
• History preservation and tamper-proofing
• Corporate transparency and accountability

Final Thoughts

The choice facing prediction markets is extremely stark: either continue to believe that humans driven by economic interests can be neutral arbiters of truth, or build a truth-determination system that completely eliminates human bias.

The answer to this question already existed—in the workings of the market itself. When $200 million was poured into a market with an obvious outcome, the “obvious answer” unexpectedly failed, exposing the system for its true nature.

The technology to solve this problem already exists.

The determination of truth is too important to be left to the highest bidder.