It has had a few updates since, then but I cannot vouch for its accuracy.
It doesn't cover audits per sé, but I feel there is important information that is tangentially related, since security audits become kind of moot if some of the items mentioned are true (i.e. CIA funding and US govt. tactics).
Full disclosure, I still use Signal for a family group chat. I have very little economic value, thus my threat model is minimal. It mentions cats several times. I neither have cats, nor interact with them frequently enough to warrant their inclusion in a threat model.
Lack of detailed audits...only in this case specifically...does not imply lack of security and/or privacy.
The protocol that Signal uses, which is in fact firmly audited with no major problematic findings, plus the fact the client is OSS is generally enough to lower any concerns.
The server side software in production for Signal.orgis not OSS. It will not be. You are required to trust the server to use Signal; because the protocol and the client renders it factually impossible for the server to spy on your messages. The server cannot read messages; or even connect who is messaging who if the correct client settings are used. (Sealed Sender).
Non-OS stats software in general is not automatically lacking in privacy or security, particularly not in this case where the affected software does interact only with software that is verifiably open-source and trustworthy in general due to the protocols and how they are implemented correctly in a verifiable manner.
Both signal and molly are considered safe, a lot of apps use the same protocol as signal, most risk come from messages leaks before the encryprion happens.
Unfortunately, I'm not aware if they did external audits, but both codes are available in github.