Infosecurity Europe: AI Coding Tools Need Built-In Security for Agentic Development Era

At Infosecurity Europe,Boaz Barzel, field CTO of Ox Security, emphasized the need for built-in security in AI coding tools to address emerging risks in an era of rapid development cycles. Traditional application security methods can't keep up with the continuous code changes enabled by AI agents. Barzel highlighted four unique attack surfaces introduced by AI: Input (instructions), Tools (servers and external connections), Execution (autonomous operations), and Output (vulnerable code).

He proposed integrating security into the coding process as a continuous, contextual function, which aims to drastically reduce vulnerability resolution times and achieve comprehensive security validation autonomously.