other discussions
 

Proton Mail’s web interface can be made to present a forged sender identity that is visually indistinguishable from a legitimate one. The problem combines three weaknesses:

Display-name text is shown in the position users read as the sender’s address. An attacker controls the From display name entirely and can put an address-shaped string there (e.g. a full name@domain.com). The interface surfaces that string prominently and does not force the actual envelope/header address into view.

The default interface font makes distinct characters identical. The UI font stack resolves to the OS system font (SF Pro on macOS), in which capital I and lowercase l render identically. A domain such as gmaiI.com (capital i) is indistinguishable from gmail.com to any reader, including a careful one. Proton has protections against several encoding-based homograph classes but not against this basic same-font case.

No authentication-failure banner is shown for sender domains without a DMARC record. When a message fails SPF from a domain that does publish DMARC, Proton displays a “failed authentication / may be spoofed” banner. When the sender domain has no DMARC record, the message is delivered to the inbox with no banner and no other flag. The inconsistency means the spoofed message arrives looking fully clean.

Taken together, a message can arrive in a Proton inbox appearing to come from a real, named person at a real domain, with no visual cue anywhere in the interface that anything is wrong. A Reply-To alias can also be set so a reply reaches an attacker-controlled address without that address being visible unless the user expands the header manually.