Linux基金会：2026开源与人工智能（AI）的未来研究报告（英文版）（26页）.pdf

1、Open Source and the Future of AIHow Agents are Disrupting Our Systems,Our Precedent,and the Human Role in SoftwareHilary Carter,The Linux FoundationAnna Hermansen,The Linux FoundationApril 2026Copyright 2026 The Linux Foundation|April 2026.This report is licensed under the Creative Commons Attributi

2、on-NoDerivatives 4.0 International Public License.Please cite this research report when using the infographics.See guidance on the reports back page.Open Source and the Future of AIThe success of open source AI infrastructure such as Ray and vLLM demonstrates three important principles:address trend

3、s,keep things simple,and remain flexible.The programmers role is evolving into an architect who designs and defines problems while delegating specific tasks and roles to neural-networked coding assistants.To build trust between individuals and the agents acting on their behalf,users need the ability

4、 to set fine-grained boundaries and privileges based on context.Without clear rules on accountability for agent behavior or a unified process for asserting identity,organizations are adopting a defensive posture that may stifle growth.While developers are moving quickly to grant agents API keys and

5、access,essential safeguards are almost entirely missing from current agent communication protocols.Reasoning traces in open models are integral to secure adoption,allowing users to inspect decision paths rather than just final output.Before an agent can automate human workflows,organizations must pr

6、ovide it with understanding by comprehensively recording processes and historical knowledge.Human accountability must remain the final stamp of quality for compliance to satisfy risk management frameworks.Open source acts against vendor lock-in and single points of failure,ensuring the flexibility t