Skip to main content
This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
MIT AI Risk Initiative
BackSpreading toxicity
Improper usage
Dangerous use

Spreading toxicity

AI Risk Atlas

IBM (2025)

Source
Sub-category

"Generative AI models might be used intentionally to generate hateful, abusive, and profane (HAP) or obscene content."

Entity: Who or what caused the harm

Human

Due to a decision or action made by humans

AI system

Due to a decision or action made by an AI system

Other

Due to some other reason or is ambiguous

Intent: Whether the harm was intentional or accidental

Intentional

Due to an expected outcome from pursuing a goal

Unintentional

Due to an unexpected outcome from pursuing a goal

Other

Without clearly specifying the intentionality

Timing: Whether the risk is pre- or post-deployment

Pre-deployment

Occurring before the AI is deployed

Post-deployment

Occurring after the AI model has been trained and deployed

Other

Without a clearly specified time of occurrence

Supporting Evidence (1)

1.
"Toxic content might negatively affect the well-being of its recipients. A model that has this potential must be properly governed."

Other risks from IBM (2025) (63)

Lack of training data transparency

6.5 Governance failure
HumanUnintentionalPre-deployment

Uncertain data provenance

6.5 Governance failure
HumanOtherPre-deployment

Data usage restrictions

7.3 Lack of capability or robustness
HumanUnintentionalPre-deployment

Data acquisition restrictions

7.3 Lack of capability or robustness
HumanUnintentionalPre-deployment

Data transfer restrictions

7.3 Lack of capability or robustness
HumanUnintentionalPre-deployment

Personal information in data

2.1 Compromise of privacy by leaking or correctly inferring sensitive information
AI systemUnintentionalPost-deployment
View all 63 risks from this paper →