There is really no way to justify theft claims in most cases, therefore something of this sort is less than feasible.  It's all subject to trust in the previously controlling figure in their claim.  At least 'loss' is an easy claim as if at any point claimed lost coins are spent, the obvious fact is they were truly not, but stolen...

Could have something on the sort of the address that held them signing a note that the coins were stolen and others signing the claim in trust, but the infrastructure for such is obviously not implemented in BTC chain.  namecoin could be used to follow such, or really any sort of other pub forum.  But either way, attempting to quarantine stolen coins in most sectors of the market is extremely problematic.  Not terrible when Gox and the like do it, but if nodes start trying to block stolen transfers, things will clearly become problematic fast.

Example of the last comment if some are not yet seeing it:  A number of coins are stolen and end up being distributed heavily before discovery of the incident, someone with even 2-3 satoshi attempts a larger transaction including these coins they now own.  If a number of nodes at this point were to say these coins were known stolen and attempt to quarantine them, the 'thieves' have already profited and network users are impacted negatively.  Further to this point, if anything managed to get implemented on a large number of nodes, someone could 'Robin Hood' and distribute a large theft across a list of public addesses immediately, before any detection algorithms could identify and validate a theft and effectively destroy the network.

The only valid solution is to safeguard owned coins, and not to attempt firewalling cures on the network.  Obviously I've gone further than the previous queries here, but worth consideration.  These concerns could be mitigated by a client implementation with wallets acknowledging a trustnet of what coins should be purged from the network, but then we're just throwing away marketable coins, which is foolish at best.