Local AI Governance: Addressing Model Safety and Policy Challenges Posed by Decentralized AI
Policies and technical safeguards for artificial intelligence (AI) governance have implicitly assumed that AI systems will continue to operate via massive power-hungry data centers operated by large companies like Google and OpenAI. However, the present cloud-based AI paradigm is being challenged by...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-07-01
|
| Series: | AI |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2673-2688/6/7/159 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Policies and technical safeguards for artificial intelligence (AI) governance have implicitly assumed that AI systems will continue to operate via massive power-hungry data centers operated by large companies like Google and OpenAI. However, the present cloud-based AI paradigm is being challenged by rapidly advancing software and hardware technologies. Open-source AI models now run on personal computers and devices, invisible to regulators and stripped of safety constraints. The capabilities of local-scale AI models now lag just months behind those of state-of-the-art proprietary models. Wider adoption of local AI promises significant benefits, such as ensuring privacy and autonomy. However, adopting local AI also threatens to undermine the current approach to AI safety. In this paper, we review how technical safeguards fail when users control the code, and regulatory frameworks cannot address decentralized systems as deployment becomes invisible. We further propose ways to harness local AI’s democratizing potential while managing its risks, aimed at guiding responsible technical development and informing community-led policy: (1) adapting technical safeguards for local AI, including content provenance tracking, configurable safe computing environments, and distributed open-source oversight; and (2) shaping AI policy for a decentralized ecosystem, including polycentric governance mechanisms, integrating community participation, and tailored safe harbors for liability. |
|---|---|
| ISSN: | 2673-2688 |