Skip to content
Search
paperApril 2026Unreviewed

Parallax: Why AI Agents That Think Must Never Act

Joel Fokou

View resourcePDF

Abstract

Autonomous AI agents are rapidly transitioning from experimental tools to operational infrastructure, with projections that 80% of enterprise applications will embed AI copilots by the end of 2026. As agents gain the ability to execute real-world actions (reading files, running commands, making network requests, modifying databases), a fundamental security gap has emerged. The dominant approach to agent safety relies on prompt-level guardrails: natural language instructions that operate at the s

Categories

Guardrails

Cite

@misc{fokou2026parallax,
  title = {{Parallax: Why AI Agents That Think Must Never Act}},
  author = {Joel Fokou},
  year = {2026},
  month = apr,
  eprint = {2604.12986},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2604.12986}
}