Algorand Foundation, the nonprofit organization behind the Algorand blockchain, announced in a press release on August 25 the public launch of AC2 (Agentic Communication and Control Protocol), an open-source, blockchain-agnostic standard for secure communication between users and AI agents. The problem AC2 targets is a genuinely practical one: as AI agents take on tasks like making payments, committing code, or authorizing purchases on a person’s behalf, there has been no standardized way for a user to approve individual actions without simply handing the agent full, unrestricted access to their accounts.
The Problem AC2 Is Trying to Solve
According to the release, the messaging channels AI agents currently rely on to communicate with users, email, Telegram, WhatsApp, were never built for this purpose. They offer no cryptographic identity verification, no standardized way to delegate specific, limited authority, and no mechanism for a user to approve one particular action without giving an agent their private key outright, the digital credential that, once shared, effectively grants total control over an account or wallet. That all-or-nothing dynamic is the specific gap Algorand Foundation says AC2 is designed to close, and it’s a problem that gets more urgent as agents take on more consequential tasks: reviewing a single suspicious email is one thing, but reviewing whether to hand an AI system standing authority over a bank account or crypto wallet is a categorically different decision, one most people have had no good tooling to make carefully.
How It Actually Works
AC2 establishes a direct, end-to-end encrypted connection, using WebRTC, a standard for real-time peer-to-peer communication, between a user’s wallet or app and an AI agent. When an agent needs to perform a signing operation, whether that’s a payment, a code commit, or an API authorization, it sends a request through AC2. The user reviews and approves that specific request through their own wallet interface, and only then is the signature delegated back to the agent for that one action. Critically, according to the release, the user’s private key never leaves their own control at any point in this process, meaning the agent gets authorization for a defined action rather than a blank check. That per-action approval model is the core design choice here: instead of a one-time decision to trust an agent broadly, a user gets a continuous checkpoint on every consequential thing the agent tries to do.
Real-World Use Cases Named in the Release
The release describes three specific workflows. A coding agent can draft code and then ask a developer to review and approve the final digital signature before it’s published, rather than committing changes autonomously. An AI assistant managing payments can send transaction details directly to a user’s digital wallet for review and approval before any payment goes through. And a shopping agent can construct what the release calls an AP2 Intent Mandate, a defined statement of exactly what it’s authorized to buy and at what price, and submit that to the user for sign-off before acting on its own. Because AC2 is built to support extensions, the release says any additional message type or signing format can be added to the protocol over time, broadening it beyond these initial examples.
The Technical Building Blocks
AC2 uses DIDComm v2.0 message formats, an existing standard for secure, interoperable messaging between different digital identity systems, intended to help AC2 work consistently across different wallets and applications rather than locking users into one vendor’s ecosystem. Authentication runs through Liquid Auth, built on FIDO2/WebAuthn, the same passkey-based authentication standard increasingly used to replace passwords on major websites, rather than a proprietary login system specific to Algorand. The protocol also supports real-time voice and text streaming alongside signing delegation, and notably requires no central message relay and no blockchain at all to operate.
Notably, No Blockchain Required
That last detail is worth underlining: despite being released by a blockchain foundation, AC2 itself is explicitly blockchain-agnostic and doesn’t require Algorand’s own blockchain, or any blockchain, to function. That’s a meaningfully different posture than a typical crypto-foundation announcement built to drive usage of a specific token or chain; AC2 reads more like an open security standard Algorand Foundation is contributing to a broader problem than a product tied to Algorand’s own ecosystem specifically.
What’s Open and What’s Not Yet Proven
The specification and reference code are available now on GitHub, alongside an open-source wallet proof-of-concept built by Pera Wallet and a developer testing plugin. Algorand Foundation says it is actively inviting design partners to review and contribute to the spec, language that signals this is an early-stage, community-input phase rather than a finished standard with established adoption. The release’s own disclaimer acknowledges that agentic commerce carries real risks, including potential fraud and identity verification issues, and recommends users and merchants confirm they’re working with verified agents and follow security best practices, a caution worth taking seriously given how new this entire category of AI-to-wallet interaction still is.
Sources
Algorand Foundation Launches AC2: An Open Protocol for Secure, User-Controlled Communication with AI Agents, PRNewswire, August 25, 2026
Editorial Disclosure
This article is based on a press release issued by Algorand Foundation on August 25, 2026, distributed via PRNewswire. Algorand Foundation is a nonprofit organization; no securities are discussed in this article, and ALGO, Algorand’s associated cryptocurrency, is a digital token rather than an equity security, so no ticker or exchange applies. Next Gen Tech Stocks was not compensated for this coverage. Statements regarding AC2’s planned features, integrations, and future protocol development are forward-looking, as the source release itself acknowledges, and are subject to change as development progresses. The release’s own disclaimer notes that agentic commerce carries risks including potential fraud and identity verification issues; this article does not independently assess AC2’s security beyond what the source release describes. This article is for informational and educational purposes only. See our full Disclaimer.
