Once RCE is confirmed, any further access from there would break ToS. It's bullshit, and kind of a cop-out - but it's a trade-off that these companies have to make. Otherwise we could employ techniques like social engineering to gain internal access, then start pen-testing from within the network.
Their ToS does not state this. They do however specifically state that social engineering attacks are not eligible.
You're actually correct. I think this situation is a bit more complicated than it seems at first.
Updated original reply.
Re:Bug Bounty Ethics
by franjkovic on Thursday December 17, 2015 @06:07PM
(#10755338)
I'd say it does.
Not interact with other accounts without the consent of their owners.
Edit: whoops I mis-read this a bit, but the point still stands - he escalated using AWS keypair that did not belong to him, and he had no consent of the owner.
Obviously that's referring to facebook accounts.
Edit: I feel that your edit is still stretching the terms a bit. It seems pretty clear that this isn't the abuse that the clause intends to prevent.
Also, the guy seems to only have verified that the credentials worked.
Except it doesn't?