The user content sandbox can be abused to trick users into opening unexpected documents after several user interactions. The content can be opened with a blob origin from the Matrix client, so it is possible for a malicious document to access user messages and secrets.
This has been fixed by https://github.com/matrix-org/matrix-react-sdk/pull/5657, which is included in 3.15.0.
There are no known workarounds.
{
    "severity": "LOW",
    "github_reviewed_at": "2021-03-02T02:47:04Z",
    "cwe_ids": [
        "CWE-345"
    ],
    "github_reviewed": true,
    "nvd_published_at": "2021-03-02T03:15:00Z"
}