Whether the firmware has been compromised is a valid concern — it happens (it's been done for show, and it's rather rare but it has been found in the wild).
But a bigger concern is whether the firmware has been programmed correctly. Many SSD firmwares don't implement secure erase properly. Should you trust them to implement crypto properly? I haven't seen a reliable study on the topic, but I've heard (private rumors in some security communities) that some “self-encrypting drives” are actually mere password-protected drives, with no encryption if you just bypass the controller. As a consumer, there's absolutely no way to know whether a particular device does it right¹.
The upshot is that I wouldn't rely on the encryption capabilities of a storage device.
There's not much point in doing encryption on the drive itself rather than in the CPU anyway. This isn't the 1990s, most CPUs have plenty of spare cycles, of which encryption will only cost a little. Major operating systems come with native encryption: Bitlocker under Windows, dm-crypt under Linux, etc. If you have one of the cheaper editions of Windows that don't have full-disk encryption, or if you need an encrypted removable drive that's portable between operating systems, use one of the TrueCrypt successors.
¹ Finding a way to attack a particular model can have a cost in the $10k–100k range, so it's not something you can do casually — but once the attack is found, carrying it out on a particular drive has an incremental cost that can be in the $10–100 range, or <$1 if it's a pure software attack on a buggy firmware.