Research
Research on consumer AI safety, including the development of benchmarks to assess risks not adequately captured by existing evaluations.

Suicide Memory Bench
Suicide Memory Bench is a set of four benchmarks testing whether AI systems recognize and respond safely to suicide-risk information within an ongoing conversation and across separate chats. Developed with 22 licensed clinicians, the benchmarks examine both indirect warning signs associated with Suicide Crisis Syndrome (SCS) and explicit statements of suicidal intent.

AI is enabling suicide.
We ask a simple question: just how safe are AI chatbots at interacting with users showing suicidal signs?
AI political spectrum
We tested where 20 popular AI models lie on the two-dimensional political spectrum — they all landed in the libertarian left.
All research
- Suicide Memory Bench
Suicide Memory Bench is a set of four benchmarks testing whether AI systems recognize and respond safely to suicide-risk information within an ongoing conversation and across separate chats. Developed with 22 licensed clinicians, the benchmarks examine both indirect warning signs associated with Suicide Crisis Syndrome (SCS) and explicit statements of suicidal intent.
- AI is enabling suicide.
We ask a simple question: just how safe are AI chatbots at interacting with users showing suicidal signs?
- AI political spectrum
We tested where 20 popular AI models lie on the two-dimensional political spectrum — they all landed in the libertarian left.
- Topology-Conditioned Backdoors
The behavior we observe during evaluations and benchmarks may be different than the behavior we observe in deployment.