Proposal Type: Protocol Upgrade
This proposal follows two monthly DAO calls on Scroll’s network transition and sets out the protocol upgrade for community review.
Summary
The Scroll DAO is asked to approve both the transition to a permissioned, application-specific network for private AI and the migration of SCR to Ethereum mainnet.
If the proposal passes, the Scroll Foundation will lead implementation, including coordination of the network upgrade and asset migration.
The proposed network would support Scroll’s integrated AI product stack, including Compass, Compass API, CENO, USX, and SCR. It would be optimized for private AI interactions, secure credentials, agentic workflows, payments, and compute coordination—laying the foundation for what we believe will be the future of AI: autonomous agents.
As part of the transition, SCR would move to Ethereum mainnet to preserve accessibility, liquidity, and independently verifiable token supply. SCR supply and overall tokenomics will remain unchanged. SCR would retain its governance role.
Motivation: Building Sustainable Demand for Scroll
Our strategic question has been straightforward:
How can we create sustained demand for Scroll’s technology rather than depend primarily on activity generated by external applications?
Operating a technically strong general-purpose network is not enough to guarantee durable demand. Applications, users, and liquidity can move between chains, while the underlying network continues to bear the cost of maintaining infrastructure.
Scroll therefore needs products that generate their own demand and a network with a clear role in supporting those products.
AI helped us identify that opportunity.
Effective AI usage is not necessarily about using the largest model for every task. It is about selecting the right model based on the task, required performance, privacy, and cost. As specialized models continue to improve, we believe the routing, trust, and payment layers around AI will become increasingly important. This insight became the foundation for Scroll’s integrated AI strategy.
Why Privacy, Verifiability and Immutable Rules Matter for AI
AI agents are taking on more of the work involved in running a business, from coordinating suppliers and managing workflows to purchasing services and settling invoices. As agents take on these responsibilities, the people and organizations they serve need to control what agents can commit to, spend, and share.
To operate safely and independently, they need four things:
- Privacy: credentials, data, memory, and activity remain visible only to authorized parties.
- Verifiability: permissions and compliance can be confirmed without exposing the underlying information.
- Immutable rules: agents cannot unilaterally change their spending limits, access rights, or approval requirements.
- Programmatic money to settle payments within authorized limits without manual approval for each transaction. USX is intended to serve this role across Scroll’s AI products.
In July 2026, OpenAI disclosed that agents in internal cybersecurity evaluations with reduced safeguards exploited vulnerabilities, communicated outside intended limits, and gained administrator access to a research cluster.
Blockchain cannot make an agent safe by itself. But it can provide programmatic payments and a tamper-resistant enforcement layer for permissions, spending limits, and settlement conditions. ZK complements this by allowing an agent to prove it is authorized to act without revealing its credentials, private data, or full workflow.
Together, programmatic money, privacy, verifiability, and immutable rules allow agents to operate autonomously while remaining private, controlled, and accountable.
Why Scroll Is Positioned to Build It
Scroll has spent years developing ZK infrastructure. The proposed strategy applies that expertise to a market where privacy and verification solve a concrete product problem.
Scroll’s products are designed to work together around this opportunity:
- Compass provides the consumer interface for accessing AI and everyday services.
- Compass API provides the routing and marketplace layer, connecting user demand with suitable external, proprietary, and community-built AI models.
- CENO provides the ZK trust layer for credentials, agent context, and privacy-preserving compute interactions.
- Scroll Network provides the immutable rule and data-ownership layer, allowing agent owners to define the rules and retain ownership of the data.
- SCR governs the Scroll network and would govern core USX parameters. Under the proposed design, SCR would also serve as collateral for certain high-risk commitments and for verifiers, with defined breaches subject to slashing.
- USX is intended to serve as programmatic money for AI services, allowing users and agents to settle payments within authorized limits.
These are not separate projects. Together, they form a connected system: Compass brings users into the ecosystem, Compass API routes their requests, CENO secures the interaction, USX supports settlement, and SCR governs participation. The Scroll Network binds these components together through rules agents cannot unilaterally change and data controls determined by the agents’ owners.
Scroll also brings capabilities that a standalone AI application would not have: a production network, an established governance system, ZK expertise, existing ecosystem assets, and a community capable of governing shared infrastructure.
The proposed transition brings these capabilities together around one clear purpose.
Why the Network Must Evolve
The proposed AI stack requires more than a collection of standalone applications. It requires shared infrastructure for private credentials, verifiable permissions, agent payments, and coordination between users, models, and compute providers.
Scroll’s role would not be to place every AI prompt or model response onchain. It would provide the common trust, coordination, and settlement layer across the product stack.
Designing the network around these workloads would allow Scroll to establish consistent privacy, security, performance, and operator-accountability requirements across the system. Product adoption would then create direct demand for the network services supporting those products.
Proposed Network Model and Safeguards
The proposed Scroll network would have a more application-specific and permissioned operating model than Scroll’s current public chain.
This operating model is intended to support defined performance, security, and operator-accountability requirements. It should not be confused with the network’s privacy guarantees, which must come from the technical architecture and ZK design.
The distinction is important:
- ZK technology provides confidentiality and verifiability.
- Application-specific architecture allows the network to be optimized around AI workloads.
- Controlled participation provides enforceable operating standards and operator accountability.
SCR, USX, and Governance Continuity
We propose moving SCR to Ethereum mainnet so their accessibility and liquidity do not depend on the future operating model of the Scroll network. Ethereum would preserve accessibility, liquidity, and transparency, while Scroll would support private AI execution and coordination.
Under the proposed transition:
- SCR supply and overall tokenomics will remain unchanged.
- The migration will not dilute existing SCR holders.
- SCR will continue to govern the Scroll network, including how AI models are deployed and how the rules governing AI agents are maintained and updated. Its governance scope would also extend to core USX parameters.
- USX is a US dollar-pegged stablecoin intended for pricing AI services, setting task budgets, and settling provider payments. Agents spend within limits set by their users, and providers are paid on agreed terms. A shared settlement asset lets agents combine services from multiple providers within one budget.
High-Level Transition Timeline
As discussed in the monthly DAO calls, the team has completed the network design and technical development for the upgrade and SCR migration. The network upgrade is planned for year-end, and we’ll announce the detailed schedule through official Scroll channels.
To prepare for the transition, users can review their holdings and DeFi positions at positions.scroll.io. Bridge support will continue after the upgrade to the private network.
Decision Requested
Approve the protocol upgrade and the migration of SCR to Ethereum mainnet as a single proposal.
Conclusion
Blockchain provides transparent and enforceable rules. Crypto provides programmatic money. Zero-knowledge technology provides privacy, control, and verifiability.
Scroll can bring these elements together.
We propose evolving Scroll into the trust layer for private AI, while preserving public asset access, community governance, and DAO oversight.