There is considerable debate over whether and how social media contributes to polarization. In a correlational study (n1 = 1,447) and two digital field experiments (n2 = 494, n3 = 1,133), we examined whether (un)following hyperpartisan social media influencers contributes to polarization and misinformation sharing. We found that incentivizing Twitter/X users to unfollow hyperpartisan social media influencers improved feelings toward the out-party by 23.5% compared to the control group, with effects persisting for at least six months. Unfollowing also led participants to engage with more accurate news and increased satisfaction with their Twitter/X feeds—without reducing engagement. This study demonstrates the long-term causal impact of exposure to hyperpartisan influencers. Moreover, unlike other social media reduction interventions, unfollowing is a targeted approach: like a scalpel, it surgically removes a few harmful parts of one’s feed, allowing the beneficial aspects to remain.