Brickken CEO: Legal liability for AI trades rests with human principals

The rise of agents capable of managing portfolios and settling payments has outpaced existing legal frameworks, leaving courts to navigate a complex mix of contract law and negligence. Because current regulations do not recognize AI as a legal person, the legal burden rests on the party that established the agent's parameters. According to Mata, an unfavorable trade result does not inherently signal a system failure; rather, it often represents a calculated risk made within the scope of an authorized strategy. Responsibility only shifts to developers or financial institutions if the agent exceeds its mandate or operates through flawed design.

To address these risks, the proposed ERC-8226 standard aims to make delegated authority verifiable onchain. By defining limits for individual transactions, asset types, and operational duration, the RAMS framework provides a clear audit trail of who granted the authority and whether the agent stayed within its bounds. While U.S. securities rules already hold broker-dealers accountable for automated market access under SEC Rule 15c3-5, industry leaders suggest that clearer standards for autonomous agents are essential to prevent legal ambiguity in the evolving digital economy.