For a trusted circle it should be easy to set up a 'middle man', an automated address that receives initial payments first, validates the intended recipient (sort of like a bank verifies the account number and name you are sending the payment to), and forwards the payment.
If the recipient is flagged by the 'community' as a fraud or criminal activity, the 'middle man' returns the payment to the sender.
I wouldn't mind building such a system, in which the community feeds data to the system to determine which addresses are trustworthy and which aren't.
But, a problem that occurs with such a system, is that the creator of the 'middle man', is able to regulate all payments within the circle. This would need an extra layer of validation, which means another Bitcoin-process within the Bitcoin-system. And with creating the 'middle man', we are actually creating a regulated system that could be taken over by authorities.
Just my 2 mBTC.
You can't use an address to identify fraud because the fraudster can use multiple addresses, what you can do is have a trust rating for an address and have people attempt to build an addresses trust rating. The "wall/circle" can be blockchain based, just a M of N sig wallet with certain functions such as escrow with timelock, you send money to the wallet and it would be returned by X date unless you and circle agree to send the money to recipient. That way the circle could protect you but not fuck you over. If the circle's "founder" where to be captured and tortured, he still could not steal the funds and at the end of X time you would recieve the funds back