New York, NY, Sept. 14, 2026 (GLOBE NEWSWIRE) -- DeepSafe, a cryptography random verification layer for cross-chain transactions, today published details of the three-stage verification pipeline its protocol runs before any cross-chain signature is produced, along with the limits of what each stage establishes.
The three stages are source-chain proof, hardware-backed authentication of the program that produced the verification result, and randomized selection of the verifiers authorized to sign.
Cross-chain security is commonly measured by validator counts and signature thresholds. Those figures describe which parties are authorized to approve a transaction, not whether the information being approved is accurate. If every validator receives the same manipulated source-chain data, raising the signature threshold does not correct the shared input.
A contract on one blockchain cannot read the state of another directly, so cross- chain systems depend on external infrastructure such as remote procedure call endpoints, light clients or relayers. An RPC response can be affected by synchronization delays, configuration errors or a compromised backend, and the exposure compounds when multiple validators query the same provider. In that configuration the signing layer can be widely distributed while the underlying data still originates from a single point of failure.
"Raising a signature threshold answers how many parties must agree before a cross-chain action proceeds," the DeepSafe team said. "It does not answer what those parties are agreeing on. The documentation is written to make that boundary explicit, including the points at which a Merkle proof or a light client cannot settle the question on its own."
Three Stages, Three Different Guarantees
DeepSafe's documentation sets out the pipeline as three questions answered in sequence: whether submitted data can be checked against evidence committed to the source chain, whether the verification result was produced by the expected program in an authenticated environment, and which validators are authorized to act on that result. Simplified payment verification, or SPV, trusted execution environment attestation, and a threshold-signing process the company calls CRVA correspond to those three stages.
At the first stage, a Bitcoin light client maintains and validates the block-header chain and uses a Merkle proof to check that a transaction is included in a particular block. Rewriting the relevant history becomes progressively more costly as additional blocks are added. Rather than accepting an RPC response at face value, the verifier checks whether the supplied data is consistent with commitments recorded by the network.
DeepSafe applies SPV as an umbrella term for its multichain light-client services, and the documentation states that the verification method varies by network. Bitcoin light clients rely on proof-of-work headers and Merkle proofs. Ethereum consensus light clients primarily use sync committee signatures to update and authenticate block headers, and verifying a specific execution-layer transaction or state value requires the corresponding data and proof. The scope and assumptions of light-client verification are therefore not identical across chains.
The documentation also separates what this stage establishes from what it does not. Inclusion of a transaction in a recognized block, consistency between submitted data and its cryptographic proof, and satisfaction of a configured confirmation requirement fall inside that scope. Whether the asset, amount and destination address match the request, and whether a message has already been executed, are determined by application logic and destination-chain state rather than by the light client. Running a full node generally provides stronger independent verification, but operating one for every supported network carries substantial storage, computing and operational cost.
Authenticating the Program Behind the Response
The second stage addresses whether a response came from the expected verification program rather than an impersonating service. DeepSafe's documentation describes a two-stage process built on Intel SGX enclaves. During initial registration, remote attestation is used to check the enclave and its code measurement, after which the corresponding public key is registered on-chain. The verification service then signs its results inside the enclave, and a Monitor component verifies those signatures against the registered key.
Attestation establishes that a response originated from a previously attested execution environment. According to the documentation, it does not establish that the program is free of software flaws, and chip security, upgrade authority and data paths outside the enclave remain within the system's trust boundary.
The documentation describes two deployment models. In Public mode, multiple Monitors can use a shared verification service, which can also be used to check the synchronization status of local instances. In Local mode, each Monitor is paired with a dedicated service, producing a more independent data path at the cost of additional deployment and maintenance work. DeepSafe's current public materials do not specify how conflicting results between the two modes would be adjudicated, or how each model is deployed across production integrations.
Verifier Selection and Threshold Approval
At the third stage, candidate nodes use Ring-VRF to prove eligibility without disclosing their long-term identity during selection, according to the DeepSafe white paper. A verifier set is chosen randomly and handles requests during the relevant epoch, which is intended to make set membership harder to predict and target in advance. Selected verifiers evaluate the request against the integration's rules, and a signature is generated once the signing threshold is met. In the lock- and-mint pattern described in the white paper, the destination-chain mint or unlock proceeds only after threshold verification succeeds.
DeepSafe has not published a single timing diagram covering every production integration, and the company states that its exact production configuration may vary by integration and that the architecture does not remove every trust assumption. Supporting references cited in the documentation include the simplified payment verification section of the Bitcoin white paper, Ethereum's light client documentation, and Bool Network, a distributed cross-chain notary platform published in IEEE Transactions on Information Forensics and Security in 2022.
"Extending this model to more networks is not a matter of adding another endpoint," the DeepSafe team said. "Each chain has its own consensus rules, finality model, proof format and data-availability assumptions, so the verification layer has to be implemented and tested chain by chain before any application logic is built on top of it."
About DeepSafe
DeepSafe is a cryptography random verification layer for cross-chain transactions. Its architecture sequences light-client verification of source-chain data, trusted execution environment attestation of the verification program, and CRVA, a threshold-signing process that uses Ring-VRF to select verifier sets randomly while keeping candidate identities undisclosed during selection. Technical documentation is available at https://docs.deepsafe.network.
Media Contact
DeepSafe
Contact: Susie Yang
Email: operation-at-deepsafe.network
Website: https://deepsafe.network
Disclaimer: The information provided in this press release is not a solicitation for investment, nor is it intended as investment advice, financial advice, or trading advice. Investing involves risk, including the potential loss of capital. It is strongly recommended you practice due diligence, including consultation with a professional financial advisor, before investing in or trading cryptocurrency and securities. Neither the media platform nor the publisher shall be held responsible for any fraudulent activities, misrepresentations, or financial losses arising from the content of this press release.
