• Frank Pasquale, Cornell Law School and Cornell TechFollow
• Vanna Carter, 2027 J.D. Candidate at Cornell Law School

Article

Fall 2025

Agentic AI, Artificial intelligence, Data protection, Automated Consumer Agents, Principal-agent law

Consumer Protection Law | Contracts | Internet Law | Privacy Law | Science and Technology Law

This essay explores how consumers might exercise newly restored agency in a digital environment increasingly shaped by generative AI. Rather than expecting consumers to read privacy-related terms of service, policymakers should encourage the emergence of Automated Consumer Agents (“ACAs”) capable of interpreting privacy provisions and acting on behalf of users according to their stated preferences. These systems could automatically reject objectionable forms of data extraction, translating consumer values into actionable decisions in online transactions.

Despite this promise, the rise of ACAs would also present new problems. Without supportive policy and legal frameworks, agentization of privacy preferences could itself reproduce or intensify existing asymmetries of power. Technology firms may develop biased or manipulative ACAs that steer consumers toward expansive data sharing while presenting themselves as neutral intermediaries. Moreover, individually rational bargains exchanging privacy for small conveniences or discounts may collectively dispossess consumers of the broader economic value generated by their data.

Even if consumers are aided by powerful ACAs, privacy governance cannot be reduced to individualized market transactions. Regulatory interventions, including anti-retaliation protections and licensing or fiduciary obligations for ACAs, are necessary. ACAs are a potentially valuable but incomplete mechanism for advancing fairness in the contemporary data economy.

Frank Pasquale, Vanna Carter, "Agentizing Privacy Preferences without Privatizing Data Protection," 39 Harvard Journal of Law & Technology 453-471 (2025)

DOWNLOADS

Since June 04, 2026