The major biogeochemical cycles that keep the present-day Earth habitable are linked by a network of feedbacks, which has led to a broadly stable chemical composition of the oceans and atmosphere over hundreds of millions of years. This includes the processes that control both the atmospheric and oceanic concentrations of oxygen. However, one notable exception to the generally well-behaved dynamics of this system is the propensity for episodes of ocean anoxia to occur and to persist for 10-10 years, these ocean anoxic events (OAEs) being particularly associated with warm 'greenhouse' climates. A powerful mechanism responsible for past OAEs was an increase in phosphorus supply to the oceans, leading to higher ocean productivity and oxygen demand in subsurface water. This can be amplified by positive feedbacks on the nutrient content of the ocean, with low oxygen promoting further release of phosphorus from ocean sediments, leading to a potentially self-sustaining condition of deoxygenation. We use a simple model for phosphorus in the ocean to explore this feedback, and to evaluate the potential for humans to bring on global-scale anoxia by enhancing P supply to the oceans. While this is not an immediate global change concern, it is a future possibility on millennial and longer time scales, when considering both phosphate rock mining and increased chemical weathering due to climate change. Ocean deoxygenation, once begun, may be self-sustaining and eventually could result in long-lasting and unpleasant consequences for the Earth's biosphere.This article is part of the themed issue 'Ocean ventilation and deoxygenation in a warming world'.