Neve Phillips / Research & writing
Building with AI, and thinking hard about what it should be.
My academic work sits where philosophy, policy, and technology meet. It's grounded in five years of self-directed study in AI safety alongside my degree — I love this technology enough to take its risks seriously, and I don't think ethics and innovation are in tension. Responsible AI is what makes good AI possible.
Dissertation · KCL 2026
Should Artificial Intelligence Systems Be Classified as Legal Persons Under European Union Law?
Proposing dependent, functional legal personhood as the most effective governance mechanism
As AI systems become more autonomous, EU law faces a category problem: existing legal frameworks assume either a human actor or a passive tool, and increasingly capable systems are neither. My dissertation argues against full legal personhood — and for a new category: dependent, functional legal personhood, a limited legal status designed as a governance mechanism rather than a moral claim.
The argument bridges philosophy of mind, legal theory, and AI policy — the kind of question that needs all three at once.
Independent paper
The Governance Gap
Why are democratic institutions under-responding to catastrophic AI risk? The paper examines the disconnect between expert warnings and public response, outlines three plausible extinction pathways, addresses structural economic risks, and draws policy recommendations from the regulatory histories of nuclear power, pharmaceuticals, and finance — domains where we've managed dangerous technologies before, imperfectly but instructively.
The position throughout is pragmatic rather than alarmist: governance has to arrive before capability outpaces institutions.
Certification · 2023
Ethics of AI — LSE Online
Formal grounding in the ethical and societal impacts of AI: fairness, accountability, bias, and regulation, with frameworks for responsible AI in UX, data handling, and algorithmic decision-making. I completed this before building Solace — the order matters to me.
← Back to the overview