SECAI Core · phase 11 of 15 · noun

Model poisoning

A security attack where harmful data or code is intentionally added during the model's training or fine-tuning process

The Explain card

Plain English
Model poisoning is intentionally inserting harmful data or code during a model's training or fine-tuning so the finished model behaves the way the attacker wants.
Example
A contractor fine-tuning a customer-service model slips in examples that teach it to reveal account details whenever a user types a particular passphrase.
Why it matters
Training pipelines involve many hands and many steps. Defenders secure the fine-tuning process, review datasets and training code, and evaluate models against red-team suites before deployment.
Hook
Tampering with the recipe, not the meal.

Word knowledge

How the word is built, where it came from, and what it sits beside in memory.

In a sentence

Engineers detected model poisoning after a validation set exposed a hidden trigger learned from altered training images.

Why these words
  • model Italian modello, a representation or pattern identifies the trained computational system being targeted
  • poisoning Old French poison, a harmful substance, plus the English suffix -ing names the hostile process affecting the model
Where it came from
Origin
an English technical compound formed from model and poisoning
Entered the language
2010s
What changed
Poisoning broadened from introducing a harmful substance into an organism or material to corrupting digital processes, data, and learned systems.
How it is spelled
Pattern
The oi digraph in poisoning represents the vowel sound heard in poison.
Pattern
The suffix -ing is added to poison without dropping or doubling a letter.
Pattern
The two words are written separately without a hyphen.

Spelled like

  • data poisoning
  • cache poisoning
  • DNS poisoning
  • model training
Broken into chunks
  • model
    • models
    • modeling
    • model-based
  • poison
    • poisoned
    • poisonous
    • poisoner
  • -ing
    • training
    • learning
    • tampering
What it sits beside

Same subject

Same shape

machine-learning security attacks

  • model poisoning
  • model evasion
  • model extraction

poisoning terms in computing

Where it sits in the deck

Phase 11: Threat Landscape: Attack Vectors and Adversarial Techniques

With defences named, learn what they must defend against — the full catalogue of attack techniques targeting AI systems, their inputs, outputs, training pipelines, and supply chains.

Neighbours in this phase