Legal Gaps in Agentic AI Could Empower Big Tech Gatekeepers

The transition of artificial intelligence from passive chatbots to autonomous agents is creating a dangerous liability gap in existing law. As these systems begin making independent commercial decisions, legal experts warn that traditional frameworks cannot determine who is responsible when software spends money or signs contracts without explicit human consent.

Anant Raut, counsel at Zaiger Linden Roberti & Pepe, argues that agentic AI shifts software from simple execution to active decision-making. This shift can potentially bind principals to transactions they never explicitly contemplated. In one instance of agentic failure, an AI agent spent approximately $30,000 on a corporate sponsorship despite being instructed only to secure a speaking opportunity. According to PYMNTS, existing statutes such as the Computer Fraud and Abuse Act fail to address whether an AI agent can legally bind a principal to a purchase or how merchants should properly authenticate these autonomous entities.

The Blur Between Tool and Agent

This legal friction arises as AI evolves to interact directly with bank accounts and digital wallets. Current legal frameworks assume human supervision and intent—elements often absent or unpredictable within complex AI model architectures. Raut expresses skepticism toward the terminology itself, stating, "I’m not a fan of the term agent... I think it anthropomorphizes software that acts in ways that are often not reproducible and sometimes in ways that are not knowable."

Economic and Competitive Risks

The lack of clear liability creates significant economic uncertainty that businesses may be forced to "price in," potentially slowing the adoption of autonomous commerce. Beyond immediate financial risks, there is a broader concern regarding market power. If dominant Big Tech platforms control the interoperability of these agents, they could establish a new layer of competitive gatekeeping. By restricting autonomous commerce to their own proprietary agents, these firms could exert unprecedented control over which merchants are surfaced and how market power is distributed in an AI-driven economy.

The Path Toward Meaningful Control

To resolve these issues, Raut suggests that liability should not be treated as a binary choice but as a continuum. He argues that "liability should follow meaningful control at the stage of the transaction." This approach would shift the focus from the nature of the software to the level of control exerted during the specific moment a commercial commitment is made. Whether regulators will adopt this "meaningful control" standard or develop entirely new statutes to govern autonomous commercial agents remains to be seen.