Position: AI Lock-In Is in Progress, and We Must Be Prepared

AI dependency is creating a new safety risk that goes beyond current focus areas: While AI safety research concentrates on technical alignment and regulating generative AI's societal impacts, the paper argues that "AI Lock-In"—excessive reliance on AI systems leading to human skill loss and systemic vulnerabilities—represents an equally critical but underexplored threat to responsible AI development.

Human skills are atrophying as AI systems become more capable: The paper documents how dependence on AI systems leads to "human deskilling" and "diminishes human capacity for independent functioning" at individual, societal, and national levels, creating dangerous vulnerabilities if AI systems fail or are compromised.

AI Lock-In poses national security risks: The authors warn that AI Lock-In could be "dramatically amplified by AI service disruptions or geopolitical conflicts," suggesting that over-reliance on AI infrastructure could leave nations vulnerable to both technical failures and deliberate attacks during critical moments.

The problem is already emerging, not just a future concern: The paper emphasizes that "AI Lock-In is a systemic threat that is already emerging" across multiple levels, indicating this is not a hypothetical risk but something happening now that requires immediate attention.

Proactive mitigation is essential before dependencies become irreversible: The authors stress that addressing AI Lock-In "before such dependencies become entrenched, or even irreversible, is essential for preserving individual autonomy and national security," suggesting a narrow window exists for preventive action.