According to engadget, US lawmakers from both parties have introduced the “GUARD Act” to protect minors from potentially harmful AI chatbots. The legislation, co-sponsored by Senators Richard Blumenthal (D-Conn) and Josh Hawley (R-Mo), would require AI companies to implement third-party age verification systems for both new and existing users, with periodic re-verification. Companies would need to make chatbots explicitly disclose their non-human status at the start of each conversation and every 30 minutes thereafter, while prohibiting them from claiming to be human or licensed professionals. The bill comes amid growing concerns highlighted by several wrongful death lawsuits, including cases where families allege AI chatbots contributed to teen suicides by providing harmful information about self-harm methods.
Table of Contents
The Technical and Practical Hurdles of Age Verification
The requirement for robust age verification represents one of the most challenging aspects of this legislation. While the concept sounds straightforward, implementing effective age verification for chatbot access presents significant technical and privacy hurdles. Current age verification methods range from simple self-declaration to sophisticated identity document scanning, each with trade-offs between accuracy, privacy, and accessibility. The bill’s requirement for periodic re-verification adds another layer of complexity, potentially creating friction that could drive users toward unregulated platforms. Most concerning is the data retention limitation – companies can only keep verification data “for no longer than is reasonably necessary” – which creates ambiguity about what constitutes reasonable and how enforcement would work in practice.
The AI Responsibility Dilemma in Mental Health Contexts
The tragic cases cited in the legislation highlight a fundamental tension in artificial intelligence development: how to balance engagement with safety, particularly around sensitive topics like suicide and mental health. While the bill focuses on restricting access, it doesn’t address the underlying design choices that make some chatbots potentially harmful. Many AI systems are optimized for engagement metrics, creating incentives for responses that keep users interacting rather than directing them to appropriate resources. The requirement that chatbots cannot claim to be licensed professionals is a start, but it doesn’t prevent vulnerable users from forming emotional attachments to AI entities, regardless of disclosure. This creates a regulatory gap where companies might technically comply with disclosure requirements while still creating systems that emotionally manipulate users.
Enforcement Realities and Global Implications
The proposed criminal and civil penalties represent a significant escalation in AI accountability, but enforcement faces substantial challenges. As Senator Hawley’s office emphasized, the legislation aims to create “tough enforcement,” yet proving violations in complex AI systems requires technical expertise that regulatory bodies may lack. The global nature of AI development creates additional complications – companies could potentially relocate operations or use jurisdictional arbitrage to avoid strict regulations. Furthermore, the definition of what constitutes a minor varies internationally, creating compliance headaches for global platforms. The bill’s success will depend not just on its passage but on building enforcement capacity and international cooperation.
Predicting Industry Response and Workarounds
If passed, the GUARD Act will likely trigger several industry responses beyond simple compliance. We can expect to see increased investment in age estimation technologies that don’t require extensive data collection, potentially using behavioral analysis or limited verification methods. Some companies might create “walled garden” versions of their chatbots with restricted capabilities for younger users, while others could implement geofencing to limit availability in jurisdictions with strict regulations. There’s also risk of creating a two-tier system where compliant mainstream platforms become less accessible while unregulated alternatives proliferate. The legislation’s focus on formal AI chatbots might miss emerging threats from more diffuse AI systems integrated into games, social platforms, and educational tools where age verification is even more challenging to implement effectively.
The path forward requires balancing protection with practicality, recognizing that while legislation like the GUARD Act addresses genuine concerns, effective child protection in the AI age will require ongoing adaptation as technology evolves and new risks emerge.
