Devcon talk will examine LLMs and bytecode exploits
Péter Szilágyi is scheduled to discuss how language models build exploits against bytecode at Devcon 8.
Devcon says Péter Szilágyi will speak about how large language models build exploits against bytecode in the Open & Verifiable Stack track. Its announcement places Devcon 8 on November 3–6 in Mumbai.
The announcement describes a forthcoming talk, not a published vulnerability or an assessment of a particular contract. Its framing raises a practical question for teams whose deployed bytecode is publicly accessible.
Developers can use the session as a prompt to revisit threat models, testing and monitoring. An audit or a tool alone cannot guarantee that a contract is safe.