[Rum] How to do the impossible: MITMs that aren't but are

[Eric Burger <[e-mail]>]

In DISPATCH it was pointed out that RUM is a man-in-the-middle (MITM) attack. We pointed out that it is a MITM that users expect.

The good news is we have existing use cases that look a lot like the RUM scenario: a PBX re-originating a call and a multimedia conference bridge. However, neither of those scenarios have end-to-end, cryptographic integrity of the identity of the end parties to each other.

I would offer that we have non-protocol solutions to this problem that could be enabled through the protocol.

How about this as an outline:
o   Note the originating party can pass an opaque token to the Interpreter to pass to the called party (e.g., a STIR token)

o   Note the Interpreter can pass that token to the called party (noting that STIR tokens are somewhat cut-and-paste attack resistant)

o   Note that in many jurisdictions, communications regulators have authority over Interpreters to ensure they act responsibly

o   Note that in many jurisdictions, communications regulators impose mechanisms that apply cryptographic tools so that a called party knows the Interpreter is really a ‘legal’ (in the legal sense) Interpreter (e.g., SHAKEN)

Would that work? Otherwise, I think we are trying to solve an unsolvable problem: specifically a MITM that knows when to set the Evil Bit.