Planetary-scale solar geoengineering interventions involve the deliberate injection of either natural or artificial particulates into the stratosphere – stratospheric aerosol injection, or SAI – with a view to offset some of the global heating caused by greenhouse gases. If implemented, the technology would create a metaphorical thermostat for the planet. Such a thermostat is advocated on the grounds that controlling global temperature reduces the harms associated with the climate crisis.

I wish to challenge this assertion.

Global temperature was first adopted at the beginning of this century as a way of indexing the extent of human impact on the climate system. Since then, managing global temperature has become the primary object of climate policy, thus the Paris agreement’s stated aim being to contain global warming between 1.5C and 2C. The policy goal of net zero emissions is derived from this target temperature range.

SAI seeks to shave off a few hundredths, or possibly two or three tenths, of a degree Celsius from this temperature index. It does so not by removing the cause of the undesired heating – the accumulation of greenhouse gases in the atmosphere – but by deliberately adding new elements, active particles, into the atmosphere. In global-average terms, this might slow the buildup of heat in the climate system; it may indeed discernibly lower the global temperature.

In my view, there are many problems with this techno-fix for the climate crisis. But I wish to draw attention here to just two: it does little to defuse most of the risks that really matter for people and ecosystems and, worse than this, it runs the risk of making some of these harms worse, perhaps even many of them. Both relate to the difference between global temperature and the everyday weather experienced by people and places.

First, global temperature is only the crudest of proxy indicators for the various and multiscale harms associated with loading the atmosphere with greenhouse gases. Among many things, these include the disruption and reconfiguration of regional weather systems, the impacts of changing extreme weather events on vulnerable local communities and ecosystems, and the consequences of the increasing acidification of the world’s oceans. Reducing global temperature by forcing it down through solar geoengineering offers no guarantees that any of these harms are reduced.

Second, artificially adding new radiatively active particles into the atmosphere will inevitably alter the dynamics of the global atmospheric circulation which determines regional and local weather. This is clear whether one draws evidence from the analogue case of explosive volcanic eruptions or from the results of climate model simulations of SAI. Hurricane tracks, the strength of the Indian monsoon, the behaviour of El Niño in the Pacific Ocean – in other words, everyone’s weather – then become things which the deliberate solar engineers must take responsibility for.

The focus of climate policy should be on interventions – whether technological, regulatory or behavioural – that deal directly with the causes of climate harms. This requires focusing on reducing emissions of greenhouse gases, implementing ways of removing them from the atmosphere, building up the resilience of human and ecological systems to a wide range of climatic hazards, and investing in the drivers of human development, which have long been shown to be cost-effective in reducing damage, morbidity and mortality from dangerous weather.

Adopting the index of global temperature as the scientific object to be controlled by SAI may give the appearance of doing some good, but this is an illusion. Pursuing SAI as a desirable and cost-effective way to reduce the primary harms of the climate crisis is an unhelpful and risky distraction from what matters most: the reduction of human emissions in the atmosphere, not the addition of new artificial materials.

• Mike Hulme is a professor of human geography and fellow and geography director of studies at Pembroke College at the University of Cambridge

• This article was amended on 20 February 2026 to correct the spelling of its author’s surname.