The Norwegian Government agreed with the leading farmers' unions to include the agricultural sector in the national effort to mitigate greenhouse gas (GHG) emissions. The parties agreed to abate 5 million t CO 2 eq in 2021–2030. Emissions shall be mitigated by reducing food waste, dietary change, and farm-level abatement measures. Among these, the farmers' unions agreed to pursue mitigation efforts at the farm level. This paper contributes to the current debate by calculating marginal farm-level abatement cost curves of seven typical Norwegian dairy farms. Dairy farms are chosen due to their contribution to the sectors' emissions. The farm-level optimization model FarmDyn is adapted to Norwegian conditions to represent local prices, yields, endowments, policies, regulations, emission calculation, and abatement technology. The model is applied to seven representative dairy farms, identified by K-medoid clustering from the Farm Accountancy Data Network. The results show that up to 14 % of farm-level emissions can be mitigated at costs below the carbon tax level proposed by the Norwegian government (2000 NOK per t CO 2 eq). Further mitigation efforts are bound to high costs. The preferred abatement measures include optimizing feeding to increase the share of concentrates, thereby reducing enteric fermentation emissions by up to 21 %. Replacing regular diesel with biodiesel and utilizing advanced manure application technology can reduce total emissions by up to 3 %. Ultimately, farms reduce their herds to mitigate further emissions, resulting in a decrease of up to 11 % and 68 % in revenues from sold milk and bull beef, respectively. This begins on farms with high stocking densities, due to their limited ability to optimize feeding. Due to high farm-level abatement costs, mitigation targets conflict with other policy goals, namely, securing farm income and maintaining production. Compensation could address the loss in income, but the reduction in production requires further action. Given the limited reduction potential for farm-level abatement at competitive costs, we suggest that dietary change and food waste reduction must achieve a significant share of the envisioned abatement target. This study provides insights into the economic feasibility of farm-level GHG mitigation by quantifying marginal abatement cost curves for Norwegian dairy farms. We highlight the financial constraints farmers could face in meeting national targets and showcase promising mitigation measures. Finally, we contribute to the current debate by demonstrating that achieving emission reductions at farm-level may compromise other policy objectives, underscoring the importance of balancing sustainability and economic viability. • First estimation of marginal GHG abatement cost curves for Norwegian dairy farms. • Adaptation of the farm model FarmDyn to Norwegian conditions. • Changes related to feeding, biodiesel and manure application have lowest costs. • High costs and conflicting policy goals question national plans to cut emissions.

Algorithm-based fertiliser recommendations offer substantial potential to improve Nitrogen Use Efficiency (NUE) and support economic and environmental sustainability. However, adoption among farmers in the United Kingdom (UK) remains limited, partly due to algorithm aversion, i.e., the tendency to distrust or avoid algorithmic-generated recommendations, even when they provide benefits. This study examines algorithm aversion in fertiliser-related decision-making among UK farmers and agronomists. Aiming to identify key barriers to adopting decision-support tools (DSTs), improving understanding of stakeholder trust dynamics, and exploring strategies to improve uptake. An online survey of 50 farmers and 26 agronomists assessed confidence in algorithmic recommendations versus human advice, understanding of NUE, perceived adoption barriers, and openness to non-traditional fertiliser recommendations. A follow-up workshop with 10 participants in DSTs trials provided qualitative insights into trust and usability. Farmers reported significantly greater trust in human advice compared to algorithmic recommendations (median 8 vs. 6, p < .001), whereas agronomists showed the reverse pattern (median 8 vs. 7.0, p < .001). Perceived barriers included cost concerns, poor system integration, complexity, and confusion over metrics. Whilst some farmers showed low levels of NUE literacy, agronomists demonstrated higher NUE literacy. Farmers relied on advice grounded in social trust and shared beliefs, while agronomists viewed algorithmic outputs as complements to technical expertise. Workshop participants found DST dashboards informative but often overwhelming. Addressing algorithm aversion through improved interface design, transparency, and tailored education, particularly via trusted advisors, may bridge the trust gap and facilitate digital tool adoption. • Farmers show significant algorithm aversion in fertiliser-related decision-making. • Agronomists are more receptive to algorithmic tools than farmers. • Perceived barriers to use include cost, poor usability, low trust, and unfamiliar metrics. • Trusted advisors may play a role in supporting farmer adoption of digital decision-support tools. • Addressing algorithm aversion has potential to improve existing on-farm practices.