It is sometimes assumed that because PGP supports older cryptographic methods that it must be at risk of something called a downgrade attack. The basic idea is that if you can't break a particular type of encryption then you trick the system into switching to something you can break. That might be no encryption at all.

When PGP is used in a unidirectional application like email a downgrade attack is impossible for the simple reason that any sort of negotiation of the method to be used is impossible. Negotiation would require some sort of reverse channel which in this case doesn't exist.

If you want to send someone an encrypted message using PGP you need one of their PGP identities. If that identity is, say, based on 2048 bit RSA encryption then you have to use 2048 bit encryption to encrypt that message. If you use any other type of encryption your correspondent will not be able to decrypt it as they will not have the corresponding private key.

Any preference information is embedded in the PGP identity (public key). This is the preference information from an identity generated from a recent version of GnuPG:

[ultimate] (1). Password Store <pass@store>
     Cipher: AES256, AES192, AES, 3DES
     Digest: SHA512, SHA384, SHA256, SHA224, SHA1
     Compression: ZLIB, BZIP2, ZIP, Uncompressed
     Features: MDC, Keyserver no-modify

The preference information is in the secure part of the identity. That means that it is signed by the public key also embedded in the identity. This is the public key that will be used to encrypt the email so the preferences are linked to the public key in a way that means that you would have to break the underlying encryption to modify the preferences. If you could do that then you would not have to bother with a downgrade attack.

PGP means that you get to choose how you want your messages sent to you. There is no opaque process somewhere in the middle that can override that preference.

PGP FAN index