In the key of C major, Cmaj7 would be I7. C7 is a bit trickier, but in almost any role it is being used in, it should be written as V7/IV. That way, it is obvious that the Bb is present rather than the B, and it also is representative of the chord's function - the V7/IV chord wants to go to the IV chord (F), where I7 (Cmaj7) is a tonic chord and very comfortable where it is. But that's only for the key of C, where C7 has an easy functional label. Roman Numeral Analysis was built for functional harmonies (see final paragraph for details), and thus it doesn't do nearly as well with non-functional chord sequences. So how would we go about doing this in F, or any other key where the two chords don't lend themselves to functional analysis?
The following workaround is not consistent with how the Roman Numeral Analysis was originally formatted, but you could also decide to always use 7 for chords with a minor seventh interval and specify maj7 if the major seventh is present (the I chord then being Imaj7, rather than just I7). This is consistent with how modern chord symbols address the issue, however.
For another potential idea, recall that much of Roman Numeral Analysis draws from the old tradition of shorthand Figured Bass notation, and the "7" symbol in Figured Bass comes from the intervals above the bass note: spelled out fully, that would be (7 5 3), but shortened to (7). Figured Bass symbols referred to diatonic notes by default, and so does RNA. When FB wanted to represent a non-diatonic note, it would be written as an accidental (there were many methods of marking chromatic alterations in FB, but let's stick to the accidental symbols for their contemporary relevance). In Figured Bass shorthand, in a minor key, (6 3#) meant to play the note which was a third above the specified bass note, but raise it one half-step, while also playing the note a 6th above the bass note.
So in F major, Cmaj7 could be written in FB as (#7) with the C specified as the bass note. Interpreted into English, this reads "play the intervals of a diatonic third, diatonic fifth, and raise the note that is a diatonic seventh above the bass by one half-step". This would not be standard in traditional Roman Numeral Analysis, but FB chromatic alteration notation applied to RNA chord inversion symbols could be a logical solution to the problem as long as it was understood correctly (and bonus points for historical accuracy).
There are still some situations in music where this would not solve the ambiguity, but be advised that if one is running into many ambiguous notation scenarios while using RNA, well... why use RNA in that case? The RNA system was built around functional harmony, and using it for nonfunctional analysis is inefficient. RNA doesn't have to explicitly encode the quality of the seventh, because the idea of reference to a key center is baked into the RNA system, allowing RNA to simply designate diatonic notes and denote other notes as alterations. In contrast to this, modern chordal symbols were designed to always be unambiguous independent of harmonic context. This is why Roman Numerals can break down outside of clear functional harmony: the RNA symbols have a hard time working around muddied or unclear key centers because they were optimized for analysis of tonal harmony.