AI Security Founder, previously safety TPM. Lover of interior design, plants, and experimental film photography.
I'm half Australian/British, and grew up bouncing between opposite sides of the globe. In high school I got into film photography, and collected upwards of 20 film cameras, along with converting a room in my parents house into a dark room. I play lead guitar (poorly), and enjoyed gymanstics, horse riding, and running growing up (I still do the latter). My favorite genre of music is UK house.
Currently: Founder of Aethra Labs - we're an AI security company using traditional cybersecurity techniques to monitor agents. We were the first startup to be part of Constellation's Incubator program.
Our runtime monitoring approach leverages system & network level observability techniques, and applies this to the problem of tamper-proof agent runtime monitoring. This approach enables all types of agents to be monitored simultaneously, without code, even third-party browsing agents. Using eBPF, we're able to monitor and intervene on any agent to prevent harmful actions.
We also provide red teaming services to AI labs and companies deploying AI products.
Previously:
I studied at the University of Oxford and University of Melbourne in Australia.
Research:
ICML 2024 - The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearning - (lead TPM)
NeurIPS 2024 - Towards Safe and Trustworthy AI Agents - (workshop; lead TPM)
2025 - The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems
Email me: izzy (at) aethra-labs (dot) com