Day 3: Davor Svetinovic - Agentic AI for Uncensorable Digital Economy | ADIA Lab Symposium 2025

Day 3: Davor Svetinovic - Agentic AI for Uncensorable Digital Economy | ADIA Lab Symposium 2025

by Davor Svetinovic

Applied Sciences & Engineering AI — News & Applications UYComputer scienceUYQArtificial intelligence (AI)
🎙 Davor Svetinovic 👥 824 📅 November 5, 2025 ⏱ 21 min 👁 152 📄 expert opinion 🧭 2026-08-16
Available in: English (current) Français

Keywords

stealth addressesnetwork neutralityMEVDeFireinforcement learning

Summary

In this talk at the ADIA Lab Symposium 2025, Professor Davor Svetinovic from Khalifa University discusses the intersection of agentic AI and blockchain technology, focusing on privacy and censorship resistance. He begins by highlighting the privacy shortcomings of major blockchains like Ethereum and Bitcoin, which lack the advances of the last decade in privacy-preserving techniques. He introduces a proposal to integrate stealth addresses into Ethereum, allowing users to receive funds without revealing the transaction to third parties, and even to the recipient until they actively search for it. This work, co-authored with Vitalik Buterin and others, aims to bring modern privacy features to existing platforms rather than creating new silos. The second part of the talk addresses network neutrality, emphasizing that privacy alone is insufficient; transactions must be includable and timely to prevent censorship. He presents a framework for understanding and mitigating censorship in blockchain systems, which is crucial for the credibility of decentralized applications. Svetinovic then connects these concepts to agentic AI, arguing that autonomous agents operating in DeFi must model latency, inclusion risks, and censorship pressures. He outlines two ongoing projects: one using deep reinforcement learning and LLMs to optimize arbitrage in DeFi, reporting preliminary improvements of 29x in simulated environments, and another on execution-aware probabilistic liquidation for Aave. He concludes that privacy and neutrality are foundational for a trustworthy decentralized economy, and that agents must be trained to navigate the extreme turbulences of crypto markets.

238 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides valuable insights into the practical challenges of implementing privacy and neutrality in existing blockchain systems, and how agentic AI can be designed to operate within these constraints. The argumentation is coherent, building from specific technical proposals (stealth addresses, censorship framework) to broader implications for autonomous agents. The speaker acknowledges the preliminary nature of some results, which adds credibility. However, the presentation is more of an overview and research agenda than a detailed technical exposition, and some claims (e.g., 29x improvement) are not yet fully validated.

Scientific Rigor, Source Quality, Title Accuracy

The talk references two recent papers, one on stealth addresses in Ethereum (co-authored with Vitalik Buterin) and another on blockchain censorship. These are credible sources, though not formally cited with URLs in the description. The title accurately reflects the content. The speaker is a recognized expert, and the content aligns with current research directions. However, the lack of specific citations in the description limits the ability to verify sources directly.

173 words

Title / Content Match

The title accurately reflects the content, which focuses on agentic AI for an uncensorable digital economy, discussing privacy and neutrality in blockchain systems.

Quality & Reliability

7/10

The talk is based on two peer-reviewed papers co-authored with leading researchers (including Vitalik Buterin) and presents ongoing research projects. However, the speaker himself cautions that some results are preliminary and not yet validated. The presentation is an expert opinion and research overview, not a formal study.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • Stealth Addresses in Ethereum (paper) — Mentioned as a recent paper co-authored with Vitalik Buterin on integrating stealth addresses into Ethereum.
  • Blockchain Censorship (paper) — Mentioned as a paper on blockchain censorship and network neutrality.

Concurring Sources

  • Ethereum Improvement Proposal (EIP) on stealth addresses — Related to the proposal for stealth addresses in Ethereum.

Contribution & Novelties

The talk contributes by framing privacy and neutrality as intertwined prerequisites for a trustworthy decentralized economy, and by proposing concrete technical and design approaches to integrate these properties into existing blockchain infrastructure. It also highlights the importance of training agentic AI to operate under these constraints, offering a research agenda for future work.

Pour aller plus loin :

  • Stealth addresses — Background on the concept of stealth addresses in cryptocurrency.
  • Network neutrality — General concept of network neutrality, relevant to the discussion of transaction inclusion.
  • Maximal extractable value (MEV) — Overview of MEV, a key topic in the talk.

99 words

Radar Profile

The radar profile shows high scores in information quality and technical level, reflecting the expert nature of the talk. The quantity of information is moderate, and the overall reliability is good but not perfect due to preliminary results. The talk is strong in providing actionable insights for researchers and practitioners.

Reliability 7/10

💬 No comments were provided for analysis.