A "threat model" in security speak is the set of adversaries you're defending against. Privacy tools aren't one-size-fits-all; what works against a curious counterparty doesn't work against a state actor.
Tier 1: Casual observers
The nosy ex, the curious co-worker, the LinkedIn stalker. They might know your wallet address from a tweet, or from an ENS lookup of your handle. They'll click into a block explorer and stop there.
Tool: A fresh stealth address per relationship. Your published meta-address gives every sender a unique one-time address; they can't link your payments together by looking at the chain. Casual observers see a meta-address but no payment graph.
What they still see: Total volume across all your stealth addresses (if they have your viewing key — which they don't).
Tier 2: Counterparties + chain-analysis firms
Anyone you transact with directly (an exchange, a merchant, a friend) plus the chain-analysis companies that sell de-anonymisation as a service. These actors have one or more known endpoints (your exchange deposit address, your merchant payment address) and try to fan out from there.
Tool: zk-rollup shielding. Aztec puts your funds in a UTXO-style shielded pool. The L1 sees your deposit; it does not see your subsequent transfers; it sees your withdrawal but cannot link it to the deposit. Chain-analysis ends at the rollup boundary.
What they still see: That you deposited into Aztec. That you withdrew. The amounts of both. Just not the intermediate hops.
Tier 3: Active network attackers
A serious adversary running their own RPC nodes, doing timing-correlation analysis, paying for VPN provider logs. They want to link your IP address to your wallet activity, your wallet activity across primitives, your real-world identity.
Tool: Penumbra. Privacy-by-default — every transaction is encrypted on-chain. Validators see only encrypted note commitments. Combined with Tor for transport (the user's responsibility, not the wallet's), this is the highest-effort privacy primitive the wallet ships.
What they still see: Tip-height growth, your transaction's existence (just not its contents).
Tier 4: State actors
A nation-state with subpoena power, the ability to issue warrants, sometimes the ability to compel exchanges/VASPs to hand over records. They can pierce most pseudonymous-chain privacy. They can sometimes pierce zk-rollup privacy if the rollup operator is in their jurisdiction. They generally cannot pierce well-implemented shielded chains with global validator distribution.
Tool: No tool. If a state actor is in your threat model, the right answer is don't use the chain at all — or use it openly and accept the consequences. Privacy tools were not designed for adversaries with subpoena power. Pretending otherwise gets people in jail.
What this means in practice
Pick a tier honestly. If you're a normal user wanting financial privacy from chain analytics (Tier 2), Aztec is your tool. If you want stealth-address discipline for receiving payments without exposing your wallet graph (Tier 1), publish a meta-address. If you have a specific Tier 3 threat (e.g. you're a journalist receiving payments from sources), Penumbra is the right tool — combined with proper OpSec.
The wallet exposes all three because different users have different threat models. The Travel Rule attestation runs across all three because MiCA doesn't care about your threat model; it cares that you've recorded the source of funds.
Next: the regulatory environment that shapes all of this.