​
Why verify signatures?

Signature verification proves your AI responses came from verified TEE hardware. This prevents tampering and ensures response authenticity. Every Confidential AI API response includes a cryptographic signature you can verify. Here we provide a Full Signature Verification Example to help you understand how to verify the signature of the response from Confidential AI API. Here we explain the key processes of verifying.

​
Get the signature

When you finish the request, parse chat_id from the response and then get the signature:
Python
...
# Parse `chat_id` from the response result
chat_id = result["chat_id"]
response_hash = result["response_hash"]
...
# Retrieve the signature
sig_response = requests.get(
    f"{BASE_URL}/v1/signature/{chat_id}?model={MODEL}",
    headers={"Authorization": f"Bearer {API_KEY}"},
)
The key fields in the signature response are:
  • text: It contains the request body hash and response body hash separated by :.
  • signature: The cryptographic signature of the text signed by the signing_address account in TEE.
  • signing_address: The public key of the account generated in TEE that signed the signature.
  • signing_algo: The signature algorithm used.

​
Verify signature

The verification process involves the following steps:
Python
message = encode_defunct(text=text)
recovered_address = Account.recover_message(message, signature=signature)
return recovered_address.lower() == signing_address.lower()
# True means the signature is valid
Or you could try verifying on Etherscan. Go to Etherscan’s Signature Verification and click Verify Signature button, then fill in the following fields:
  • Address: Use the signing_address from the attestation API
  • Message: Use the text field from the signature response
  • Signature Hash: Use the signature field from the signature response
Click Continue to see the signature verification result.

​
How signature verification works

The signature proves two things:
  1. Request integrity: The text field contains hashes of both your request and the AI’s response
  2. Hardware authenticity: The signature comes from the TEE’s signing_address you verified earlier
When you verify the signature, you confirm:
  • The response matches what the TEE generated
  • No one tampered with the data in transit
  • The response came from genuine TEE hardware

​
Verify using Etherscan (optional)

You can also verify signatures visually using Etherscan Tool:
  1. Enter the signing_address from your attestation report
  2. Paste the text field from the signature response
  3. Enter the signature value
  4. Click verify to see the confirmation
This gives you a visual confirmation that everything matches correctly.
Signature verification completes your trust chain. You’ve now verified the hardware, software, and response authenticity - your AI workload is fully secured!