Fintech & Crypto Alerts · Cameron Ellis · 18 November 2026

Bitcoin covenants Part 6: How FE covenants restrict keys

Bitcoin covenants Part 6: How FE covenants restrict keys

FE covenants encode Bitcoin spending constraints into the signing key itself, so only transactions that meet a predefined condition can be signed. Cointelegraph Research’s bitcoin covenants Part 6 on covenants explains Jeremy Rubin’s 2024 FE’d Up Covenants design: nodes still verify with OP_CHECKSIG, avoiding a soft fork—but the cryptography remains impractical today.

Key Takeaways

For readers following Fintech & Crypto Alerts, the latest research chapter reframes a long-running Bitcoin debate: can covenants be enforced without rewriting consensus?

What are FE covenants and how do they work?

Functional encryption (FE) generalizes public-key cryptography. Instead of a private key that can sign any message, a functional key only yields a chosen function of a secret. In Rubin’s design, that function is a signing routine: it outputs a Bitcoin signature only for allowed transactions.

According to Cointelegraph Research, the function does not sign with the base private key directly. It tweaks the key with a hash of the spending policy, so a violating transaction can at best produce a signature for a different public key that does not control the coins.

Setup requires an authority that generates a Bitcoin key pair and an FE key pair, publishes the public material and functional key, then must destroy the secret material. Addresses can commit to a 2-of-2 script pairing a user-controlled key with the FE signature so outsiders cannot freely trigger permitted spends.

Why would FE covenants matter for Bitcoin?

Changing Bitcoin consensus is hard: miners, node operators, developers and economic actors must coordinate. That friction has pushed researchers to seek covenant-like behavior without soft forks. FE covenants keep verification identical to today’s OP_CHECKSIG path while encoding the spending policy in the key.

A general FE construction that takes a program and a zero-knowledge proof can be more expressive than OP_CTV-style single-transaction covenants. The proof can also let policy depend on off-chain data, while keeping the fixed signing function itself smaller to evaluate.

Are FE-based Bitcoin covenants practical yet?

Not yet. Multi-input FE of the kind needed is tightly linked to indistinguishability obfuscation. A 2017 experiment needed 3.7 hours to obfuscate a tiny 12-bit circuit, used about 120 GB of disk, and took 67 minutes plus 128 GB of RAM to evaluate—still “far from practical,” the researchers said.

Practical FE schemes today cover narrow functions such as inner products, not Bitcoin ECDSA signing. Rubin’s paper lists under-developed cryptography and the setup trust assumption as core downsides. CTV-style opcodes remain cheaper in bytes and verification but need a base-layer change. Cointelegraph Research says the next chapter covers Un-FE’d Covenants, which swap FE for oracle signers bonded through BitVM.

← Open in blast feed