AI accountability: building secure software in the age of automation (www.techradar.com)

AI tools are rapidly reshaping software development—42% of developers report at least half their codebase is AI-generated—but that productivity surge is amplifying security risk. Studies cited in the piece show a sharp rise in malicious activity (58% of orgs seeing more AI-powered attacks), doubled network traffic for a third of organizations, and a 17% year-on-year breach increase. Development teams are alarmed: 80% express security concerns about AI-assisted coding, especially around outdated or insecure third‑party libraries, hidden vulnerabilities, and overreliance on models whose internal logic developers may not fully understand. Gartner even predicts many GenAI proofs-of-concept will be abandoned by end-2025 due to poor security controls. The article urges a secure-by-design mindset: bake security into every phase, validate AI outputs rigorously, and maintain foundational practices (input validation, least privilege, threat modelling). Practical steps include vetting and policy-testing AI tools by security teams, continuous developer upskilling, clear governance that’s seamless to follow, and controlling data (including local hosting of models when appropriate). Key implications for the AI/ML community are clear: balancing efficiency with accountability will determine whether AI-enriched development yields resilient, ethical systems or accelerates new attack vectors and systemic failures.