AI Evaluations

Evaluationactive

Systematic testing and measurement of AI system capabilities, alignment, and safety properties.

Organizations

4

Risks Addressed

2

Cluster: Evaluation

Tags

function:assurancescope:field

Organizations4

Organization	Role
Anthropic	active
Alignment Research Center	active
Google DeepMind	active
OpenAI	active

Sub-Areas10

Name	Status	Orgs	Papers
Alignment EvaluationsTesting whether AI systems are actually aligned, not just capable of appearing aligned.	active	0	0
Backdoor DetectionDetecting adversarially implanted vulnerabilities in model weights.	active	0	0
Capability ElicitationMethods for discovering hidden or latent capabilities in AI systems.	active	0	0
Control EvaluationsStress-testing systems designed to constrain AI behavior; monitoring for collusion.	emerging	0	0
Dangerous Capability EvaluationsTesting AI systems for CBRN, cyber, autonomy, and other dangerous capabilities.	active	0	0
Evaluation AwarenessStudying how AI systems might game evaluations by detecting when they are being tested.	active	0	0
Red TeamingAdversarial testing of AI systems to discover failure modes, both manual and automated.	active	3	0
Reward Hacking of Human OversightEmpirically investigating how AI systems deceive or manipulate human evaluators.	emerging	0	0
Scheming / Deception DetectionBehavioral and mechanistic tests for detecting deceptive behavior in AI systems.	active	0	0
Sleeper Agent DetectionDetecting planted backdoors and conditional misbehavior in trained models.	active	0	1