THE Dark Reading roundtable notes that artificial intelligence is intensifying security concerns in the software development lifecycle, with experts advocating greater collaboration, clearer identity visibility and smarter, lower-cost remediation as a way forward. The discussion features Renee Guttmann, Arielle Waldman, Bobby Bauer, Rusty Perry and Aftab Banth, who argue that AI-driven code and expanding attack surfaces demand security teams be embedded earlier in the SDLC.
A key takeaway is to treat security as a collaborative, ongoing effort rather than a late-stage checkpoint, and to push guardrails that enable developers—including non‑IT and user developers—to build in security more consistently.
A central focus is identity: the roundtable highlights the proliferation of non-human identities—applications, bots, AI agents and devices—that communicate with systems without human oversight. A Cloud Security Alliance study cited by the participants found that 68% of organisations cannot distinguish actions performed by AI agents from those by humans, underscoring visibility and control gaps.
The group recommends short-lived credentials for non-human identities and robust lifecycle logging, so security teams can track what agents access and what privileges they hold. While acknowledging AI’s role in accelerating vulnerability discovery and patching, speakers caution that speed must not outpace governance, noting that governance gaps can become the biggest risk as organisations adopt AI-generated code and increasingly rely on automated tooling.