RoboTech Founder Warns AI Cannot Be Trusted Without Proof

Denis Saklakov, founder and managing partner of RoboTech Frontier Hub, argues that artificial intelligence systems cannot be trusted in financial markets unless their decisions are recorded on tamper-proof infrastructure. In an interview published on September 17, Saklakov laid out a case for blockchain-based verification as the missing accountability layer for autonomous AI agents. The argument lands at a moment when AI accountability is under intense scrutiny. OpenAI disclosed six cases of misaligned model behavior a day earlier, and regulators across the European Union and the United States are weighing new rules for autonomous AI systems that manage money, approve loans, or execute trades without human sign-off.

Blockchain as a Compliance Ledger, Not a Compute Layer

Saklakov, who holds a master's degree from Northwestern University and completed MIT's Applied AI Science postgraduate program, draws a distinction between running AI models on-chain and using blockchain strictly as a record-keeping layer. The accelerator he founded in 2024 develops projects spanning AI verification, robotics, and human-machine interfaces, including an AI investment tool called Meijin that monitors portfolio assets and manages exit strategies based on investor risk profiles."Blockchain can record the relevant state of the system, permissions, decision conditions and execution history," Saklakov told crypto.news. He emphasized that the value lies in "provenance and a record that is difficult for the decision-making system itself to rewrite," rather than in making AI models inherently more accurate.That framing positions blockchain not as a tool for improving AI outputs but as an external auditor that logs what an AI system decided, when, and under what parameters. If a trading bot executes an order that later draws regulatory attention, an on-chain log would provide a timestamped, immutable trail that neither the operator nor the model provider could alter after the fact.

Meijin and the Case for Deterministic Logic

RoboTech's own Meijin tool illustrates the philosophy. The tool uses AI for analysis but relies on deterministic, auditable logic for final execution, rather than allowing a language model to improvise the final decision. It monitors cryptocurrency portfolios around the clock. The tool tracks assets after purchase and triggers exit strategies based on pre-set risk parameters, producing a decision trail that can be reviewed or audited without reverse-engineering a neural network's reasoning. The distinction matters because most AI-powered financial tools today operate as black boxes. When a model hallucinates a data point or executes a trade based on fabricated inputs, there is often no accessible record of the internal state that led to that action. Saklakov argues that blockchain provides the external checkpoint that internal model logs, which the model developer controls, cannot credibly offer.

Regulators Are Already Moving Toward Audit Requirements

The European Union's AI Act began phased enforcement in 2025, with logging requirements for high-risk AI systems applying once the relevant high-risk rules take effect. The United States Securities and Exchange Commission (SEC) has identified automated investment tools, AI, and trading algorithms as areas for examination in its FY2026 priorities Neither framework mandates blockchain specifically, but both demand the kind of tamper-evident, third-party-verifiable record-keeping that distributed ledgers are designed to provide.Whether financial institutions adopt blockchain-based audit trails will likely depend less on the technology's capabilities and more on whether regulators accept on-chain logs as compliant evidence. The SEC's FY2026 examination priorities, which cover automated investment tools, AI, and trading algorithms, will help shape how regulators assess these systems.