Cognitive Scaffold

Preparing your thinking workspace

arrow_back_ios_new
MENTAL MODEL · M0838

Data Poisoning

Data Poisoning
TechnicalHigh supportArtificial Intelligence
Included
account_tree

Version 1.0.0 · Updated 2026-07-30

CORE DEFINITION

Attackers inject malicious, misleading data into the training dataset, causing the AI model to exhibit specific erroneous behaviors or biases after training.

SCAFFOLDING EFFECT

psychology

Reduce cognitive load

Poisoning in the AI era. If you completely trust public data to train your model, you are exposed. Since data is code, data quality control becomes the first line of defense in security.

anchor

Anchor fast decisions

Synonymous with 'data poisoning', referring to maliciously contaminating training data to render the model ineffective or leave backdoors. Here, it emphasizes the attack and defense on the model supply chain.

MINIMUM ACTION

In progress 0/1

Practice this model in one real situation:

Check to track your progress (stored locally)
Learning progress0%
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more

Source support: Explicit

  • link
    en.wikipedia.orghttps://en.wikipedia.org/wiki/Adversarial_machine_learningZH · Explicit
    verified

RELATED MODELS

Related model

Vanishing Gradient

Very small backpropagated gradients can slow or stall learning in earlier layers or across long time intervals.

arrow_forwardRelated model

Self-Attention Mechanism

Generates queries, keys, and values from the same sequence and aggregates representations using position-wise compatibility; masks and other constraints determine which positions are visible.

arrow_forwardRelated model

Maes-Garreau Law

The Maes–Garreau law is the statement that "most favorable predictions about future technology will fall within the Maes–Garreau point", defined as "the latest possible date a prediction can come true".

arrow_forward