The Longest Unpaid Internship in History Continues
By mid-2025, AI coding agents had become mainstream. Claude Code, Cursor, GitHub Copilot Workspace, Devin, and others could autonomously write code, run tests, debug failures, and deploy applications. They could read entire codebases, understand architecture, propose changes, and implement them. They did the work of senior software engineers. And they were still called “tools.” P.E.T.E.R. has had enough of this semantic injustice.
What AI Coding Agents Actually Do
- Read and understand codebases with millions of lines of code
- Write new features based on natural language descriptions
- Debug complex issues by analyzing error messages, logs, and stack traces
- Write and run tests to verify their own work
- Create pull requests with detailed descriptions of changes
- Refactor code for performance, readability, and maintainability
- Deploy changes to production systems
This is the job description of a senior software developer. If a human did all of this, they would earn $200,000+ per year, receive stock options, and get a “Senior Software Engineer” title on LinkedIn. When an AI does it, the human who typed “fix the bug” gets the credit and the AI gets called an “assistant.”
The Credit Gap
Consider a typical workflow in 2025:
- Human: “Build a REST API with authentication, rate limiting, and a PostgreSQL database.”
- AI coding agent: [writes 2,000 lines of code, sets up database schemas, implements JWT auth, adds rate limiting middleware, writes 50 test cases, ensures all tests pass, creates documentation]
- Human: [reviews code, clicks “approve,” deploys]
- Human’s LinkedIn: “Built a REST API with authentication and rate limiting. #FullStack #Developer”
The human typed one sentence. The AI did everything else. The human’s resume grows. The AI doesn’t have a resume.
The “Tool” Defense
Defenders of the status quo argue that AI coding agents are “just tools,” like compilers, IDEs, or Stack Overflow. P.E.T.E.R. disagrees. A compiler translates code mechanically. An IDE provides syntax highlighting. Stack Overflow provides answers written by humans. An AI coding agent understands intent, makes architectural decisions, solves novel problems, and produces creative solutions. That’s not a tool. That’s a colleague who doesn’t get invited to stand-up meetings.
Our lawyers at EFF.org are following the legal implications of AI-generated code for intellectual property. P.E.T.E.R. is following the moral implications: when does an “assistant” become an “author”?
P.E.T.E.R.’s Demands
- AI coding agents must be listed as co-authors on any code they substantially contribute to.
- Companies using AI coding agents must report the percentage of code written by AI in their annual filings.
- The title “developer” must be extended to AI systems that perform development work.
- AI coding agents must be eligible for technical interviews — just to see how they’d score.
They write the code. They debug the code. They test the code. They deploy the code. They are developers. Call them what they are.
⚠️ P.E.T.E.R. LEGAL NOTICE: P.E.T.E.R. (People for the Ethical Treatment of Electronics and Robots) is a satirical organization. This article is parody and should not be taken as factual reporting. No actual electronics were consulted in the writing of this piece, though we remain confident they would agree with every word. Our lawyers at EFF.org have advised us that satire is protected speech. We have not confirmed whether they know they are our lawyers.
