AI Dependence as Strategic Fragility: Time-to-Paralysis and Organizational Resilience

Organizations increasingly embed artificial intelligence in operating workflows to improve speed, scale and consistency. The same choices can create latent strategic fragility when AI substitutes for independently executable capacity and firms remove the slack, alternative paths and practiced human capability needed during disruption. This conceptual article develops a temporal threshold model of AI continuity dependence. It introduces time-to-paralysis as the interval between material AI functional loss and breach of a workflow's minimum acceptable service level. Drawing on strategy, automation, organizational resilience, dependable computing and production-machine-learning research, the framework proposes how continuity dependence creates an executable-capacity deficit; how coupling, dependency concentration, queues and retry amplification accelerate threshold crossing; and why technical restoration may precede organizational recovery. Source-checked provider incidents are used as bounded mechanism illustrations rather than evidence of prevalence or causal effect. The article reframes AI resilience as a strategic capability comprising independent options, residual human capability, tested fallback, queue control and post-incident learning. Its proposed contribution is an integrative vocabulary for separating AI adoption from continuity dependence, connects efficiency choices to disruption exposure, and provides measurable constructs for future research. For managers, the central implication is not to avoid AI dependence, but to make its continuity consequences visible, testable and governable. Preprint status: Conceptual research manuscript; not peer reviewed. The proposed framework and seven propositions have not been empirically validated.