SDL Says No to AI-Generated Code as the Open-Source Debate Heats Up

A lot of people are talking about using AI in software development, and one of the most popular open-source libraries for games has made its point clear.

The Simple DirectMedia Layer (SDL) is a key part of making games that work on more than one platform and the Steam Runtime ecosystem. It has now officially said that it will not accept code contributions made by AI or large language models (LLMs).

This action shows that the open-source community is becoming more divided over whether or not AI should be used to make software.

Growing Use of AI in Development

SDL's decision comes at a time when more and more developers are using AI-powered coding tools like ChatGPT, Claude, GitHub Copilot, and Grok.

These tools say they will:

• Speed up development

• Take care of repetitive tasks

• Make it easier to write complex code

But the people who keep SDL going are fighting against this trend because they are worried about the code quality, the legal risks, and the project's long-term integrity.

The Policy: No AI-Generated Code

At the heart of SDL's new policy is a clear and strict rule:

Contributors are not allowed to use AI or LLMs to write code for the project.

This includes any code that was written with the help of tools like ChatGPT or Copilot.

Developers who want to add to SDL must:

• Show that they wrote the code themselves

• Ensure it follows the project’s licensing rules

Legal and Licensing Concerns

There are both legal and technical reasons for this policy.

One of the biggest problems is that it's hard to tell where AI-generated code comes from. Large language models learn from huge datasets that include code that anyone can see, but some of this code may be subject to strict licenses.

This creates risks such as:

• Accidental license violations

• Legal exposure for the project

• Conflicts with SDL’s Zlib license

The people who run SDL are worried that accepting these kinds of contributions could get the project into trouble with the law.

Reliability and Trust Issues

In addition to licensing issues, there are also worries about how reliable code written by AI is.

SDL developers have said that AI tools often give results that are wrong or misleading. These systems are known to “hallucinate” problems, meaning they:

• Identify issues that don’t exist

• Suggest fixes that don’t work in real-world scenarios

Because of this, it is hard for maintainers to trust code that hasn't been fully understood and written by a person.

Limited Role of AI

SDL is not completely against using AI as a tool.

Developers can use AI to:

• Look for bugs

• Analyze code

However:

• All fixes and final code must be written by humans

This highlights SDL’s position:

AI can assist, but cannot replace human responsibility.

What Triggered the Decision

One reason for this policy's introduction was recent activity within the SDL project.

• Some code reviews had used tools like GitHub Copilot

• There were rumors of contributions made using LLMs like Claude Code

The maintainers were concerned, leading them to formalize a strict no-AI policy.

A Divided Open-Source Ecosystem

SDL's decision fits into a broader trend where open-source projects are taking different approaches to AI.

• SDL → Complete ban on AI-generated code

• Linux kernel community → Allows AI with disclosure and accountability

This difference shows that there is no clear consensus on how to handle AI in open-source development.

The Accountability Problem

One of the main reasons this debate is happening is accountability.

Open-source contributors are expected to:

• Fully understand their code

• Take responsibility for it

With AI-generated code, this becomes unclear:

• Can developers truly verify correctness?

• Who is accountable for errors or violations?

Many believe this challenges the core principles of open-source development.

Rise of “AI Slop”

Another concern is the growing number of low-quality AI-generated contributions, often referred to as “AI slop.”

This leads to:

• Poor-quality submissions

• Increased review workload

• Slower progress for maintainers

A Protective and Philosophical Move

SDL's policy can be seen as both:

• A protective measure against legal and quality risks

• A statement of values emphasizing human authorship

The goal is to:

• Maintain high-quality code

• Ensure transparency

• Avoid legal complications

Questions for the Future

The decision raises important questions:

• Will more projects enforce strict bans?

• Will AI make such bans harder to maintain?

• Could these rules slow down innovation?

AI-assisted development can significantly boost productivity, especially for smaller teams. However, avoiding it may impact speed.

Speed vs Trust Debate

Speed (AI Adoption)	Trust (Human Code)
Faster development	Higher reliability
More automation	Clear accountability
Rapid innovation	Legal clarity

Supporters of SDL argue that:

• Trust and reliability matter more than speed

• Bugs or legal risks can have serious consequences

Conclusion

In the end, SDL's stance highlights a major challenge in the tech industry: balancing innovation with responsibility.

As AI continues to reshape software development, decisions like this will influence how the broader ecosystem evolves.

SDL’s refusal of AI-generated code is a significant step in this ongoing debate, prioritizing:

• Human authorship

• Legal clarity

• Code reliability

The industry’s response will determine whether this approach becomes the norm or remains an exception.