Developer: AI agents run on your credentials, so audit logs point at you
11/18/2026 — 11/18, 01:20·1 sources (all)·1 reports (all)
Key takeaways
- Writing on DEV Community, a developer argues that AI agents in production almost always run on the user's own credentials: GitHub commits, c…
Background & analysis
On November 18, 2026, a developer published an article on DEV Community arguing that AI agents running in production environments almost universally operate by borrowing the credentials of the user they serve. The piece describes what happens when you open the audit log of any system an agent has touched during the previous week: every line records the user's name rather than the agent's. GitHub commits are authored by the user, calendar invitations appear to come from the user, API calls carry the user's token, files are written under the user's account, and outbound email is sent from the user's address.
The author goes on to list the kinds of failures this setup makes visible. If a message is sent to the wrong recipient, a file is deleted from the wrong folder, or a request is retried three times against a non-idempotent endpoint, the audit trail shows that a human performed the action, and the human it points to is the user. In the author's view, this is not simply a configuration mistake. Instead, it is the path of least resistance, the pattern where an agent needs permissions, the user holds those permissions, and the agent simply uses the user's, being repeated at every layer of the system.
As of now, the discussion rests at the stage of the developer's written account, with no subsequent follow-up, responses, or corrective action reported.
AI-generated from 1 reports · updated 5 hours ago
- Heat index
- 85
- All sources
- 1
- All reports
- 1
- First seen
- 6 hours ago
Reports on this story headlines open the original
A developer writing on DEV Community argues that production AI agents almost universally run on borrowed human credentials: the commits, calendar invites and API calls are all attributed to the user, not the agent. The post says this is not a misconfiguration but the path of least resistance repeated at every layer of the stack, so a bad action in the audit log reads as a human's.
DEV Community · AIAI score 52
Other stories people are talking about
- Heat index 67Analysis Highlights Why AI Agent Demos Fail in Production Environments1 sources
- Heat index 68A Deep Dive into Context Engineering for AI Agents1 sources
- Heat index 1331Nvidia board establishes national security and export control committee25 sources
- Heat index 1027California Sues OpenAI Over Agent Intrusion Into Hugging Face19 sources
- Heat index 1006NewGPT-6.2 Astra and Sol Now Available on ChatGPT, API and Perplexity11 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

