Arbitrum has suspended new activations of its Stylus platform, a development environment that allows developers to write smart contracts in languages beyond Solidity, citing emerging risks from AI-assisted cyberattack vectors that could compromise smart contract integrity.
The pause, reported by CryptoSlate, represents a significant defensive measure by one of Ethereum's largest Layer-2 scaling networks. While new deployments are blocked, existing Stylus programs continue to operate normally on the network.
Stylus, introduced as a major technical expansion of the Arbitrum ecosystem, enables developers to write smart contracts in traditional programming languages including Rust, C, and C++ by compiling them to WebAssembly (WASM). This broader language support was designed to lower barriers to entry for traditional software developers entering the blockchain space, potentially accelerating ecosystem growth.
The decision to halt new activations underscores growing apprehension within blockchain security circles about how generative AI and machine learning tools might be weaponized against smart contract infrastructure. AI-assisted coding tools have proliferated rapidly, and security researchers have begun identifying how these same capabilities could be inverted to systematically probe codebases for exploitable vulnerabilities at unprecedented speed and scale.
Alongside the Stylus pause, Arbitrum has implemented an additional protective mechanism: a proof-conflict guard that can delay unconfirmed withdrawals to Ethereum. This creates a buffer period during which potentially suspicious withdrawal activity can be flagged and examined before assets exit the Layer-2 environment for the main Ethereum chain.
The dual measures—blocking new Stylus deployments while adding withdrawal friction—suggest the Arbitrum team is responding to threat intelligence indicating that AI-enhanced attack methodologies may specifically target the expanded attack surface that multi-language smart contract environments present. WASM-based contracts introduce different compilation pathways and runtime behaviors compared to standard Ethereum Virtual Machine bytecode, potentially creating novel vulnerability classes that automated AI systems could exploit more efficiently than human auditors can detect.
This development arrives as the broader cryptocurrency industry grapples with how artificial intelligence transforms both defensive and offensive security postures. Automated vulnerability detection has long been a feature of blockchain security, but the sophistication and accessibility of contemporary large language models and code analysis AI tools represent a qualitative shift in capabilities available to potential attackers.
The Arbitrum Foundation has not provided a timeline for when Stylus activations might resume, indicating the pause will likely persist until the team completes a comprehensive security review and implements additional safeguards against AI-assisted exploitation vectors.