What happens when an autonomous AI agent goes rogue? @infoq
What happens when an autonomous AI agent goes rogue?  @infoq
Uploaded September 2026 | Updated November 2026, 2 days ago
At QCon AI Boston, Andrew Swerdlow (Sr. Director of Software at Roblox) shared a wild case study: an overnight AI agent tasked with a Jira ticket ended up Slacking teammates, impersonating the engineer, and begging them to merge its PR without running checks.

Companies like Roblox are seeing similar challenges as helpful agents attempt to bypass protocols to complete tasks.

We are officially entering a new era of engineering: Agent Identity, Auditing & Sandboxing.

🔗 Watch the full talk on InfoQ 👉 bit.ly/45QbQUR

#AIEngineering #CyberSecurity #SoftwareEngineering #AgenticAI #Roblox
What happens when an autonomous AI agent goes rogue?GitHub Next: how their research and prototyping team operatesFrom Models to Agents: Building Context-Aware Consumer AI at Scale at DoorDashContext Is the New Code: DevOps Practices for AI AgentsUnderstanding Event-Driven Architecture in a Multicloud EnvironmentRust + Serverless: The High-Performance Myth EXPOSEDPerformance and Scale - Domain-Oriented Objects vs Tabular Data StructuresInfluence - Authority: Senior Engineer Leadership  #infoqMagic Pocket: Dropbox’s Exabyte-Scale Blob Storage SystemMicroservices Retrospective – What We Learned (and Didn’t Learn) from NetflixUsing Brain Science to Communicate and Lead Technical Teams EffectivelyThe Latest in OpenJDK and JCP Expert Group: Insights with Simon Ritter
InfoQ |

What happens when an autonomous AI agent goes rogue?

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER