Impulse response functions (IRF), the response in a climate parameter to an emission pulse of CO2, are used to characterize Earth system response timescales and to calculate Global Warming Potentials (GWPs). GWPs are widely used to compare emissions of different greenhouse gases and to compute CO2 equivalent emissions as reported by governments to the United Nations Framework Convention on Climate Change (UNFCCC). The GWP of any gas x is the absolute GWP of gas x Absolute and relative Global Warming Potential (AGWPx) divided by AGWP of CO2. Ideally, AGWP and GWPx would be independent of atmospheric CO2 and climate. However, AGWP , and, in turn, GWPx change under rising atmospheric CO2 and global warming, affecting the emission reporting under the UNFCCC. Here, we apply perturbed parameter ensemble simulations, constrained in a Bayesian approach by observational data, to investigate how AGWP and IRF vary under different atmospheric background CO2 levels (CO ). We provide analytical formulations to compute AGWP and IRF for CO2, ocean and land carbon uptake, global mean surface air temperature, steric sea level, and ocean heat content, and to adjust these metrics to different CO . AGWP , given by the time-integrated response in CO2 at year 100 multiplied by its radiative efficiency, is 101.8(±13.5) 10−15 yr W m−2 kg-CO for CO = 425 ppm and decreases by 7% for CO = 500 ppm. The decrease is driven by a decrease in the radiative efficiency of CO2, partly canceled by a concomitant increase of IRF due to muted ocean and land carbon uptake under higher CO2 levels. We recommend regularly adjusting AGWP and, in turn, GWPs of long-lived gases to contemporary atmospheric CO2 and climate.