The first attack involves an intermediary node losing funds while another party gains funds, allowing an attacker to extract funds from the intermediary node and ultimately syphon all of the low-reputation (LOW-REP) node's funds. However, Jager assumes that high-reputation (HIGH-REP) nodes will not collaborate with attackers and this proposal can work in practice. The second attack involves an attacker node receiving an HTLC from a victim and charging hold fees without forwarding the payment. Just before the timeout for the HTLC would force settlement on-chain, the attacker removes the HTLC without forwarding it or tries to forward it at the last moment to blame someone else for its failure. This allows the attacker to extract the maximum hold fee without downstream hold fees cutting into their profits. Jager notes that this attack requires an untrusted attacker node to charge hold fees, but does not jeopardize the reputation of the HIGH-REP node that collaborates with the attacker.