liveBook · Manning

Prompt injection

As explained in Machines that Think

An attack that embeds instructions in input content to manipulate a language model processing that content.

Prompt injection embeds instructions in content that a language model processes. In peer review, concealed instructions may be unnoticed by human reviewers yet influence AI-assisted assessment. section 8.4.3

Concealed instructions in papers

Instructions in white text or tiny fonts may try to influence AI-assisted peer review, making submitted papers part of an adversarial setting. section 8.4.3

How Machines that Think covers it

How deep it goes
Defines the attack through concealed text and its possible effect, without detailing defenses or implementation.
Examples
Uses hidden white text and tiny-font instructions in submitted papers as examples.
What you can do after
Evaluate how concealed instructions can turn AI-assisted review into an adversarial process.
How it connects
Connects language models with peer review and research-integrity safeguards.

Also explained in

Compare how 16 books explain prompt injection →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Machines that Think. The section links point to the parts of the book it draws on.