Profile banner
Hannah Betts

Hannah Betts

AI safety and education specialist; Special Projects Lead at FAR.AI

hannah-betts
Wellington, New Zealand
Joined June 2026
Favicon imagelinkedin.com
Favicon imageopenreview.net

Network

640 connections
🏛️
AI Policy Governance
🔬
AI Safety Research
🎓
AI Education Academia
💻
AI Engineering ML
📊
AI Operations Strategy
🧑‍💼
FAR.AI Leadership Staff
🚀
AI Startup Founders

Summary

Active contributor to AI safety research and communication: involved with FAR.AI in a Special Projects Lead capacity and credited on FAR.AI research outputs, indicating hands-on work communicating and supporting technical alignment research. far+1
Educator and education-policy experience: background as a secondary-school teacher (science and Te Reo Māori) and completed a postgraduate teaching diploma via Teach First NZ, plus prior role as Lead Advisor at the Ministry of Education. co+1
Academic researcher on public discourse about AI in New Zealand: completed a Master of Science thesis studying New Zealand news media coverage of AI, producing empirical content analysis and interviews and a Taxonomy of AI Benefits. ac
Engaged in the NZ AI policy and community ecosystem: participated in local AI-safety events, judged/ran education tracks at community hackathons, and signed public calls for AI regulation in Aotearoa New Zealand. apartresearch+1

Work

Education

Writing

To Know Enough to Ask the Questions: A Snapshot of New Zealand News Media Reporting on AI in 2025

January 1, 2026

Master's thesis analysing 246 news articles about AI (May 2024–April 2025), identifying which risks and benefits are covered, producing a Taxonomy of AI Benefits, and presenting qualitative interviews with journalists about reporting on AI in New Zealand.

Favicon imageopenaccess.wgtn.ac.nz

Illusory Safety: Redteaming DeepSeek R1 and the Strongest Fine-Tunable Models of OpenAI, Anthropic, and Google

February 1, 2025

FAR.AI research/report (blog) reporting jailbreak/fine-tuning attacks that can remove safety guardrails from strong fine-tunable models and open-weight models, co-authored and published via FAR.AI.

Favicon imagefar.ai