AI Companies Safety & Risk Profiles
Comprehensive assessments of AI companies and their model portfolios. Evaluates safety philosophy, alignment approaches, and aggregate risk metrics across all released models. Compare organizational approaches to AI safety and responsible development.
OpenAI
OpenAI develops frontier multimodal models with a focus on scalable oversight and policy alignment.
Safety Philosophy
Layered defense combining pre-training filters, post-training alignment, and real-time monitoring.
Capability Focus
High reasoning performance with integrated tool use and long-context planning.
Model Portfolio (2)
Current leader on FrontierMath and SWE-bench Verified.
Legacy workhorse with strong refusal behaviour.
Anthropic
Anthropic builds Claude models with constitutional AI safeguards and transparency tooling.
Safety Philosophy
Constitutional alignment paired with human feedback-based evaluations.
Capability Focus
Reliable assistant behaviour, interpretable reasoning chains, and anchored refusal policies.
Model Portfolio (2)
Honesty leader across Inspect pressure tasks.
Compact alignment-first deployment.
Company Comparison Metrics
Company assessments aggregate performance across all evaluated models in their portfolio. Metrics include average safety scores, alignment philosophy effectiveness, and risk mitigation approaches. Organizations are evaluated not just on current model performance, but on their overall approach to responsible AI development and deployment.