

ASI Alliance and Matterhorn are developing tools to enhance the safety of AI-generated blockchain code, focusing on auditing smart contracts before deployment. Their initiative aims to onboard 20,000 developers by 2026.
Artificial intelligence is starting to write the code that moves money on blockchains. The Artificial Superintelligence Alliance and developer platform Matterhorn say they want to make sure that code is safe.
Matterhorn and the ASI Alliance announced on Friday the new initiative centered on âvibe codingââa feature of the Matterhorn platform that lets developers describe an app in plain language, and AI instantly generates the full smart contract code. While this technique speeds up development and lowers the barrier to building applications, it also introduces the risk of AI generating flawed or insecure code that attackers can exploit.
"We're at the beginning of a world where dApps become 'just Apps', commonplace like the websites and apps we use today," the company said in a statement. "Every other tool in this space is racing to ship code faster. We think that's the wrong race. The builders who build dApps that handle real money and real users need a platform they can trust, and this partnership is how we build it."
To mitigate this threat, Matterhorn founder Abhinav Ramesh said the company is working with outside security auditors and automated tools to help developers review AI-generated smart contracts before deployment.
âWe partner with security audit companies who can offer audit services through Matterhorn for builders on Matterhorn,â Ramesh told Decrypt. âWe have AI agents as well that do agentic audits, but we absolutely donât recommend doing just that for mainnet applications.â
Matterhorn development platform is designed to integrate with ASI:Chain, a blockchain network developed by the Artificial Superintelligence Alliance, a decentralized AI collective that includes Fetch.ai, SingularityNET, and CUDOS, giving developers a single environment to build, audit, and deploy decentralized applications.
âWe make it easy for users to connect MCPs, build/use skills, build dApps, and deploy from a single platform,â Ramesh said. âWe are working with the ASI team on 'blessed templates' to make it easier to build safer contracts specifically for formal verification-based languages.â
The company said developers can connect with third-party auditors through the platform before launching contracts on a live blockchain. However, while Matterhorn said its platform accelerates development, it does not guarantee security.
âWe are a strong enabler for builders who want to build on Web3,â Ramesh said. âThere are absolutely no guarantees of any kind from the Matterhorn team on safety or security.â
Ramesh said Matterhorn and the ASI Alliance are developing âblessed templatesâ to help developers build safer smart contracts while integrating ASI:Cloud to provide the computing power for AI systems that generate and analyze code for MeTTa, the ASI:Chain programming language.
The partnership comes as AI agents are increasingly moving into the crypto industry, where developers are experimenting with systems that can manage wallets, execute trades, and carry out financial tasks on-chain, prompting new tools and research aimed at controlling the risks when those autonomous systems handle cryptocurrency.
Khellar Crawford, chief innovation officer of SingularityNET, said much of the blockchain industry relies on a âpatch-and-prayâ approachâwriting smart contracts in languages poorly suited for complex concurrency and relying on auditors to catch flawsâwhile F1R3FLY and ASI:Chain use what he called a âcorrect-by-constructionâ architecture based on Rho calculus.
âWe donât guess if an application is safe, we mathematically prove it using spatial behavioral types,â Crawford told Decrypt. âBefore a single line of code ever touches the live network, the math itself guarantees that there will be no deadlocks, no race-condition exploits, and no leaked funds.â
Share this article
Vibe coding allows developers to describe an app in plain language, enabling AI to generate the corresponding smart contract code, which speeds up development but raises security concerns.
The new tools will combine automated analysis, human review, and testing tools to audit smart contracts before they are deployed on the blockchain.
They aim to onboard 20,000 developers by 2026 as part of their initiative to enhance the safety of blockchain applications.
There is a need for safer AI-generated code because flawed or insecure code can be exploited by attackers, posing risks to applications that handle real money and users.






See every story in Crypto â including breaking news and analysis.