What If Your AI Agent Is the Attacker? | Online Webinar
The next DeFi hack won't need a bug in your smart contract. It just needs one injected prompt.
Join Stephen Ajayi, Leading Offensive Security Engineer at Hacken, for a hands-on technical session on how prompt injection attacks hijack AI agent toolchains in Web3 — and the defense architecture that actually stops them.
From One Prompt to Drained Wallet: The Full Kill Chain
We walk through how a malicious instruction enters an AI agent's context and becomes a signed, irreversible on-chain transaction — across 5 stages: Inject → Hijack → Abuse → Build TX → Broadcast.
What We Cover
Attack surfaces unique to Web3: token metadata, governance proposals, on-chain event logs, IPFS/Arweave, price feed labels, TX memo fields
6 real attack patterns: direct wallet drain, indirect injection, RAG poisoning, cross-agent hijacking, payload splitting, multimodal injection
Defense-in-depth: 7 layers from input filtering to anomaly detection — including the Transaction Policy Gateway that turns a compromised agent from a catastrophic loss into a blocked attempt
Real-world cases: Drift Protocol ($285M) and Resolv Protocol ($23M) — what went wrong and what agent builders must learn
Who Should Attend
Web3 developers and protocol architects integrating AI agents
Smart contract auditors expanding into AI security
Security leads and CTOs shipping autonomous agents on-chain
Speaker: Stephen Ajayi, Leading Offensive Security Engineer, Hacken
Format: Technical Presentation + Q&A
About Hacken
Hacken is an end-to-end blockchain security and compliance partner for digital assets. Born on blockchain in 2017, Hacken combines deep DLT expertise with enterprise-grade quality, AI-powered offensive security, and globally recognized standards. Trusted by 1,500+ adopters — including the European Commission, ADGM, MetaMask, Ethereum Foundation, and Binance — Hacken delivers provable assurance for digital-asset systems across security, transparency, and regulatory readiness.