Sunday, 22 November 2026
Bitcoin Covenants: Functional Encryption Moves Rules to Signing Keys

Image: © Cointelegraph

Fintech & AIWednesday, November 18, 2026 at 07:06 PM UTC

Bitcoin Covenants: Functional Encryption Moves Rules to Signing Keys

Researchers propose using functional encryption to enforce Bitcoin spending constraints within signing keys, potentially avoiding the need for consensus-level soft forks.

By Salala Times Desk
Key points
  • Jeremy Rubin proposed using functional encryption to move covenant logic from Bitcoin Script to the signing key in a May 2024 paper.
  • This approach allows nodes to verify signatures using standard OP_CHECKSIG, removing the requirement for a base-layer soft fork.
  • The method relies on a trusted setup authority to generate keys and must destroy the master secret to prevent unauthorized spending.
  • Current functional encryption implementations are computationally expensive and far from practical for general Bitcoin use cases.

To address this, researchers have explored methods to enforce covenant behaviors without altering the underlying protocol. A new technical approach proposes embedding spending constraints directly into the cryptographic signing key, allowing the network to validate transactions without a soft fork.

This concept, detailed in a paper titled FE’d Up Covenants, was introduced by Jeremy Rubin, the original author of the OP_CTV proposal.

How Functional Encryption Works

In traditional public-key cryptography, a private key possesses the ability to sign any message. The proposed functional encryption model restricts this capability, ensuring that a key can only sign transactions that satisfy a predefined set of conditions. Bitcoin nodes would continue to verify these signatures using the existing OP_CHECKSIG operation, treating them as standard valid signatures.

Functional encryption generalizes public-key encryption by issuing subsidiary functional keys derived from a master secret. These keys reveal only specific functions of the plaintext rather than the full message. For Bitcoin applications, the function in question is a signing operation. The system requires multi-input functional encryption because the signing function depends on both the private key and the transaction data being signed.

Trust Assumptions and Setup

The implementation of this covenant system requires a setup authority to generate the initial key pairs. Crucially, the authority must be trusted to destroy the master secret and the original private key after the setup phase. If this trust is compromised, the authority could sign spends that violate the covenant’s spending policy.

Once the setup is complete, the system generates dedicated key pairs for each covenant by tweaking the original keys with a hash of the spending policy. The receiving address typically commits to a 2-of-2 script, requiring a signature from both the user and the functional encryption system. This structure prevents unauthorized parties from triggering transactions that might otherwise be permitted by the covenant logic.

Practical Limitations

Current constructions require computational resources that exceed current technological capabilities. The cryptographic primitives involved are at least as complex as indistinguishability obfuscation, which remains largely at a toy scale in implementation.

  • Researchers spent 3.7 hours to obfuscate a small 12-bit pseudorandom function, a circuit far simpler than ECDSA signing.
  • The obfuscated program required 120 GB of disk space and 128 GB of RAM for evaluation.
  • Practical functional encryption schemes currently cover only narrow function classes, excluding the complex signing required for Bitcoin.

Future Directions

Looking ahead, alternative approaches are being developed to approximate covenant behavior without relying on impractical functional encryption. One such method involves replacing functional encryption with oracle signers bonded through BitVM. Additionally, other proposals have shifted from functional encryption to witness encryption to achieve similar goals.

Sources
Based on reporting by Cointelegraph.

More in Fintech & AI

View all →