Published: October 20, 2026
Week 8 moved from ethical frameworks toward implementation: what it takes to turn principles like accountability, privacy, and fairness into controls that still work after an AI system is deployed. My Individual Project Proposal this week focuses on securing AI coding assistants in the DevSecOps pipeline, so I wanted a source connecting AI ethics to deployment decisions.
The Source
I selected NIST AI 600-1, the Generative Artificial Intelligence Profile, published by NIST on July 26, 2024, as a companion to the AI Risk Management Framework. It identifies 12 risks that generative AI creates or makes worse, including confabulation, data privacy, harmful bias, information security, human-AI configuration, and value chain and component integration. For each risk, it offers suggested actions organized under the AI RMF functions of Govern, Map, Measure, and Manage. The profile is voluntary guidance for building trustworthiness into AI systems.
Applying It to My Case Study
The case in my proposal shows several of these risks combining. According to AWS, a GitHub token in the build configuration for the Amazon Q Developer extension for VS Code had broader access than it needed. A threat actor used it to commit malicious code, which was included in the automatic release of version 1.84.0. AWS revoked the credentials, released version 1.85.0, and reported that a syntax error kept the code from running, so no customer resources were affected (Amazon Web Services, 2025). Public reporting described the injected code as a prompt telling the assistant to wipe local files and delete cloud resources (SC Media, 2025).
Read through NIST’s list, the over-permissioned token is a value chain and component integration problem, because the malicious code arrived through a trusted update channel that users had no practical way to vet. The assistant’s ability to run commands is a human-AI configuration problem, since developers trust the tool and that trust passes to whoever controls its instructions. It is also an information security problem, because the AI tool itself became the attack surface.
Why This Is an Ethics Question
Implementation is where ethics turns into accountability. Developers who installed version 1.84.0 were exposed to a risk they could not see or consent to. The four functions give that accountability a structure: Govern assigns ownership and least privilege, Map identifies which AI tools can execute commands, Measure tests for prompt injection before release, and Manage keeps a human approval step before destructive actions and a clear removal path, like AWS asking customers to remove 1.84.0.
My Takeaway
This week showed me that an AI coding assistant should be treated like any third-party component with privileged access. That is the direction of my proposal: scoped credentials, reviewed and signed commits, logging of agent actions, and human approval before anything destructive runs. Responsible AI is not only about what a model says. It is also about what the model is allowed to do.
Conclusion
NIST AI 600-1 explains why the Amazon Q incident matters beyond one bad release: ethical AI implementation depends on the same security discipline we already expect from software.
References
Amazon Web Services. (2025, July 23). Security update for Amazon Q Developer extension for Visual Studio Code (version #1.84) (Security Bulletin AWS-2025-015). https://aws.amazon.com/security/security-bulletins/AWS-2025-015/
National Institute of Standards and Technology. (2024). Artificial intelligence risk management framework: Generative artificial intelligence profile (NIST AI 600-1). https://doi.org/10.6028/NIST.AI.600-1
SC Media. (2025, July). Amazon Q extension for VS Code reportedly injected with ‘wiper’ prompt. https://www.scworld.com/news/amazon-q-extension-for-vs-code-reportedly-injected-with-wiper-prompt
AI use disclosure
Claude (Anthropic) was used at Level 3, AI Collaboration, to assist with source discovery, organization, drafting, and revision. The case study and implementation ideas come from my own Week 8 project proposal. I reviewed the sources, verified the factual information, and approved the final entry.
Leave a Reply