Applying Existing Legal and Governance Frameworks to AI-Influenced Decision-Making.
The governance challenge posed by artificial intelligence is institutional, not technological. The critical question is not what AI does, but who decided to use it.
Executive Summary
Artificial intelligence (AI) is increasingly used across government, business, and defence. Public debate often frames AI as an existential threat, raising concerns that autonomous systems may undermine traditional structures of responsibility.
This paper argues that the governance challenge posed by artificial intelligence is not primarily technological but institutional.
Legal systems already contain well-established doctrines for assigning responsibility when harm arises from complex chains of action involving multiple actors.
The real risk lies not in a disappearance of responsibility, but in failures of organisational governance. Where responsibility appears unclear, it is often because institutions have not clearly allocated ownership, oversight, or accountability for the deployment of AI systems.
Effective AI governance therefore requires institutions to ensure that responsibility for decisions influenced by artificial intelligence remains clearly defined, documented, and subject to appropriate oversight.
Artificial intelligence may influence decisions, but it does not change who remains responsible for them.
Key Findings:
- AI does not remove human responsibility for decisions. Legal responsibility continues to rest with identifiable actors and institutions.
- Existing legal doctrines – including negligence, corporate liability, product liability and duty of care – are capable of addressing harms involving AI systems.
- The idea of a “responsibility gap” is often overstated. Apparent gaps usually arise from unclear governance structures rather than a lack of legal accountability.
- The real challenge for organisations is governance: ensuring that responsibility for the deployment and supervision of AI systems is clearly allocated and documented.
- Effective AI governance therefore requires institutional accountability rather than entirely new legal frameworks.
- Effective governance must focus on decision accountability rather than technology control.