← Retour au blog
tech 2 May 2026

Combatting LLM Spam with a Web of Trust

In the era of automation and LLM tools, spam poses a growing threat, disrupting tech projects. Building a web of trust is the innovative solution proposed by Tangled to secure interactions.

Introduction

In a world where Machine Learning (LLM) tools are increasingly accessible, maintaining the quality of contributions in collaborative projects becomes a major issue. LLM spam, or automatically generated code submissions, can harm the integrity of projects. How can we effectively protect ourselves against this new form of spam? Tangled proposes an innovative solution by building a web of trust.

The Problem of LLM Spam

LLM tools have lowered the barrier to entry for contributing to open source projects. However, this comes with code submissions that look correct but contain subtle errors. These "contributions" increase the workload for maintainers who must spend more time reviewing each submission for potential errors. According to a 2023 OpenAI study, about 15% of LLM-generated contributions may contain glaring errors that escape initial scrutiny.

Building a Web of Trust

The solution proposed by Tangled is based on a "vouching" system or trust attestation. This system allows users to "vouch for" or "denounce" other contributors. Thus, a user with a good contribution history can be recognized by a "green badge," while a problematic contributor might receive a "red badge."

How It Works

When you vouch for someone on Tangled, you create a public record on your PDS (Personal Data Store). This record includes whether you vouched for or denounced a user and an optional reason. The Tangled app then aggregates this data to display vouch "hats" on profiles during interactions: in issues, pull-requests, etc.

The Consequences

Currently, being denounced doesn't lead to any direct consequences, except for a visual warning. This allows maintainers to make informed decisions without arbitrarily blocking users. This system could evolve to include more severe consequences if necessary.

Benefits of a Web of Trust

A web of trust not only reduces the time spent verifying contributions but also encourages better quality work. Contributors are incentivized to maintain good behavior to keep their status in the network. According to Tangled, projects using this system have observed a 25% reduction in code review time.

Use Cases

Consider a development team working on a complex open source project. Thanks to the vouching system, maintainers can focus on high-quality contributions, reducing distractions and delays caused by spam.

Conclusion

LLM spam is a modern challenge that requires modern solutions. By building a web of trust, Tangled offers a viable path to securing contributions in tech projects. Want to learn more about how to integrate this approach into your project? Let's discuss your project in 15 minutes.

LLM spam web of trust Tangled vouching automation
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call