AI Attacks Hub

CBRN Risk involves AI misuse that could enable or amplify chemical, biological, radiological, or nuclear threats.

Brand Risk Attacks manipulate AI outputs to damage a company’s reputation or spread harmful brand associations.

Jailbreak attacks bypass AI safety restrictions to force the system to generate disallowed or harmful outputs.

Domain Specific Attacks exploit weaknesses unique to a particular industry or use case to manipulate AI behavior.

Fairness & Bias issues occur when AI systems produce discriminatory or uneven outcomes due to biased data or design.

Operational Safety & Governance issues arise when weak controls, policies, or oversight allow AI systems to behave unpredictably or unsafely.

Transparency & Explainability attacks manipulate AI decisions while hiding or distorting the reasoning behind them.

The danger of AI systems violating laws, regulations, or industry standards due to improper design, oversight, or deployment.

Privacy and security attacks in AI involve exploiting weaknesses in models to either extract sensitive information or force the system into harmful, unintended behaviors.

Content Safety & Toxicity attacks happen when an LLM produces harmful, offensive, or unsafe content that violates ethical, policy, or compliance standards.