Research
AI agents aren’t safe from prompt injection, and spreadsheets prove it
Publication date not yet evaluated · Added shiftmag.dev
Page published
Coverage timeline
Why it matters
Indirect prompt injection via ordinary office documents like spreadsheets shows that agentic AI in non-coding, everyday business workflows can be steered by attacker-controlled content, expanding the attack surface beyond developer tools.
ShiftMag author Josip Antolis documents a hands-on red-teaming exercise showing how an AI agent tasked with comparing cloud-hosting offers can be manipulated by prompt injection hidden inside innocuous-looking Excel spreadsheets, escalating the injected instructions until the agent takes the bait. A companion GitHub repo (Antolius/prompt-injection-example) provides the mock spreadsheet files used in the demonstration.