# Algorand Launches AC2 to Secure AI Agent Interactions Without Exposing User Keys The new protocol enables secure AI agent operations while keeping cryptographic credentials fully isolated. - **Source:** [DailyCoin](https://brifing.si/source/3e9b0f85-cb2f-40ee-ba61-db9410cf0937) - **Published:** 2026-08-26 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/fd28a7ef-e21c-49c6-a933-a06d5bc3ec8a ## Core Purpose of AC2 Protocol Algorand Foundation unveiled AC2, an open protocol allowing AI agents to request sensitive operations without holding users' private keys or API credentials. Launched on August 25, this open-source solution ensures no transaction occurs without explicit user consent. The protocol combines open standards including DIDComm v2.0, WebAuthn/FIDO2, and WebRTC DataChannels for secure communication. ## Risks in Current Systems When managing payments or authorizing digital operations, most chat-based approval systems lack cryptographic proof of intent. Credentials typically reside in the agent's runtime environment, creating theft risks if compromised. AC2 aims to eliminate this security gap by implementing hardware-bound authentication for every transaction request. ## Doorbell Analogy Operation AC2 functions like a doorbell: AI agents request access, users approve via their own devices, and keys never leave user control. Instead of credential sharing, each operation requires separate approval through hardware-bound FIDO2 passkey signatures. This approach prevents unauthorized access even if the agent's environment is breached. ## Technical Architecture Details The protocol integrates three core technologies: DIDComm v2.0 for message formatting, WebAuthn/FIDO2 for phishing-resistant authentication, and WebRTC DataChannels for peer-to-peer communication post-handshake. It operates without centralized relays or blockchain dependency, though supporting on-chain use cases like x402 payments remains possible. ## Integration Simplicity Algorand emphasizes that basic AC2 implementation requires only a plugin rather than full-stack overhaul. The system supports payment approvals, git commit signing, API access authorization, and bounded delegation scenarios where agents operate within strictly defined limits. Reference implementations include wallet pairing tools and framework plugins. ## User-Centric Control Philosophy Foundation executives position AC2 as enabling useful agent functionality without granting authority to act against user interests. The open-source release includes GitHub repositories, reference implementations, and wallet integration tools. The protocol maintains credential isolation while providing verifiable proof of intent for each agent request. ## Industry Impact Potential AC2 adoption hinges on whether agent platforms view hardware-bound approvals as practical security upgrades or unnecessary friction. For developers building agentic commerce tools, it offers human-in-the-loop control emphasizing credential protection. The protocol could set new standards for autonomous system security in financial and development contexts. ## New Benchmark in Digital Security AC2's launch potentially establishes new security paradigms at the AI-blockchain intersection. By enabling secure AI interactions without key exposure, the protocol may become a milestone in digital asset management. Its applications in financial transactions and code signing could revolutionize how autonomous systems handle sensitive operations while maintaining user sovereignty. ## Sources covering this story - https://dailycoin.com/algorand-launches-ac2-to-keep-ai-agents-off-users-keys/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._