Cinchor has launched a system that records AI agent decisions on a blockchain to prevent tampering. When an AI agent makes a decision—such as approving a refund—the record's cryptographic fingerprint is written to the Omne blockchain at that moment, making it impossible for the company or operator to change the record later without the proof failing. Users can verify decisions themselves against the blockchain without needing to trust the company's internal logs.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Cinchor, a startup, has built a system that records AI agent decisions (like a support agent approving a refund) on the Omne blockchain—an append-only chain outside the application's control—so that changing any character in the record after the fact breaks the cryptographic proof.
Why it matters
Unlike internal audit logs that rely on a company's own database, this approach lets anyone verify an AI decision was actually made as recorded, without trusting the company. A user can check the blockchain proof themselves; if the record is tampered with, the proof fails. This matters for businesses that need to demonstrate to customers or regulators that their AI agents acted according to stated rules and did not alter the record afterward.
What to watch
The system is available now with no signup required, and Cinchor invites users to verify the example (a refund approval) themselves or issue and verify their own records on the chain.
Cinchor has released a system for recording AI agent decisions in a tamper-proof way. The example on the company's page shows a support agent that auto-approved a refund under a stated policy ceiling. At the instant the agent made that decision, Cinchor wrote the record's fingerprint to the Omne chain, an append-only blockchain outside the application's control. The key innovation is that users need not trust Cinchor's word: they can verify the proof themselves by checking the record against the on-chain fingerprint. If someone tries to change any character in the record afterward, the cryptographic proof fails, exposing the tampering. This differs fundamentally from an internal audit log, which relies on the company's own database and requires the user to trust the company's assurance that the log is accurate and unaltered. Because the fingerprint is committed to the blockchain at the moment the decision is made, neither the operator nor Cinchor can rewrite the agent's action and still pass verification. Cinchor invites users to try the system without signing up, verify the example refund approval themselves, or issue and verify their own records on the chain. The company provides code for verification and documents its full trust model, including concessions.
Cinchor addresses a fundamental trust problem in AI systems: proving that an agent's decision was made as claimed and has not been altered. Traditional audit logs live inside a company's own database, so a user must trust the company's word that the record is accurate and unchanged. By committing a cryptographic fingerprint to the Omne blockchain—an append-only ledger outside the application's reach—at the moment a decision is made, Cinchor creates third-party evidence. The blockchain's immutability means that any subsequent tampering with the record breaks the proof, making the dishonesty detectable. This shifts accountability from self-attestation to verifiable, chain-backed proof.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
