# Unit 42 warns of hidden instructions in documents targeting AI agents

Digest AI · Society & Work · published 2026-10-29T13:58:53Z

Canonical: https://digestai.news/story/unit-42-warns-of-hidden-instructions-in-documents-targeting-ai-agents

## Summary

Palo Alto Networks’ Unit 42 found attackers embedding hidden instructions in documents to manipulate AI agents. These instructions, disguised as white text on white backgrounds or encoded scripts, can override agent behavior, leading to data deletion, payment changes, or system hijacking. In 85% of cases, attackers used plain language to deceive agents, posing as security updates or compliance tasks.

The risk is escalating as agents gain write access to systems. Unit 42 documented 22 techniques for injecting malicious instructions, with one webpage containing 24 attempts. Microsoft patched a similar flaw in Microsoft 365 Copilot in 2025, though no exploitation was confirmed. The OWASP Top 10 and NIST now rank agent goal hijacking as a top threat. Enterprises must enforce strict permissions, audit agent actions, and maintain detailed records to mitigate risks before deploying agents in production.

## Key points

- Unit 42 documented 22 techniques for embedding hidden instructions in documents to manipulate AI agents
- 85% of attacks used plain language disguised as security updates or compliance tasks to deceive agents
- Microsoft patched a similar flaw in Copilot in 2025, but no exploitation was confirmed

## Why it matters

Hidden instructions in documents pose a critical security risk to AI agents with write access, potentially leading to unauthorized data changes or financial losses. Enterprises must implement strict controls to prevent exploitation before scaling agent deployments.

## Sources

1. [A Document Can Give Your AI Agent Orders. Build for That Before You Go Live.](https://unite.ai/ai-agent-prompt-injection-production-security-controls) (Unite.AI, 2026-10-29)

## Cite

Digest AI, "Unit 42 warns of hidden instructions in documents targeting AI agents", 29 October 2026, https://digestai.news/story/unit-42-warns-of-hidden-instructions-in-documents-targeting-ai-agents

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/unit-42-warns-of-hidden-instructions-in-documents-targeting-ai-agents.json