50+
PhD level evaluation and red teaming experts at Faculty
3 labs
Evaluation and red teaming work with Google, Anthropic and OpenAI
CBRNE
Misuse evaluations, plus systemic harms such as manipulation and radicalisation
01 / 03
Red teaming and evaluation of generative AI in an enterprise HR setting, alongside safety evaluation of Gemini.
- Accenture
Accenture Responsible AI
Google HR red teaming and evaluations
Red teaming and evaluation of generative AI used in Google’s HR processes, led by Ted Kwartler, Managing Director, Responsible AI at Accenture.
- Gemini 4
Faculty
Gemini 4 Safety & Security Report
Listed by Faculty among the frontier model safety reports connected to its evaluation and red teaming work.
02 / 03
Anthropic
ClaudeBiosecurity evaluation design for Claude 4, and Claude as the engine of our rapid evaluation.
- 2025
Faculty
Claude 4 biosecurity evaluations
Faculty helped design the biosecurity evaluations for Anthropic’s Claude Opus 4 and Sonnet 4 models, documented in the Claude 4 system card.
- 2025
Faculty
Claude Opus 4.5 system card
Faculty cites the Claude Opus 4.5 system card among the frontier model reports tied to its evaluation work.
- Today
Accenture and Faculty
Claude Fable rapid evaluation
Claude Fable writes, runs and grades the test suites in our two week rapid evaluation; Faculty scientists verify high severity findings.
03 / 03
OpenAI
GPTRed teaming before launch of OpenAI’s o1 reasoning model.
- 2024
Faculty
o1 red teaming before launch
Faculty red teamed OpenAI’s o1 model before launch, as documented in the o1 system card.
“It’s absolutely paramount that foundation models are built safely.”
The same method, applied to your AI.
Sources: Faculty, AI evaluations and red teaming and the model system cards it links. Partner names and marks belong to their owners.