Skip to content
Exclusive study 2025

AI in Programming 2025

Global use of AI in programming: fact-based, critical and free of marketing bias.

130+
Sources
84%
Adoption
41%
AI code
As of December 2025
100% free
No spam
150+ page PDF (German)
The study

Facts instead of marketing bias

A critical look at the most profound upheaval in the history of software development

This study analyses one of the most profound upheavals in the history of software development. In less than three years, AI-powered programming tools have gone from experiment to global standard.

84 percent of developers worldwide use AI tools. 41 percent of global code is already written with AI assistance.

The analysis draws on more than 130 international sources: developer surveys, market research and academic studies. It shows how strongly AI already shapes programming, productivity, code quality and the security situation today.

The results paint a nuanced picture. Alongside the opportunities, there are considerable risks:

  • Productivity gains depend heavily on context and cannot be generalised
  • Security vulnerabilities in automatically generated code occur systematically
  • Experienced developers are losing trust in the accuracy of the tools

This study offers a realistic assessment beyond marketing promises. It shows how companies and developers can use AI sensibly, responsibly and strategically.

Key findings

The most important insights

A preview of the full results

84%

of developers worldwide use AI-powered programming tools

41%

of the world's code is already written with AI assistance

45%

security vulnerability rate in AI-generated code

33%

of developers see AI as a potential threat to their jobs

Contents

What the study covers

6 chapters of in-depth analysis and concrete recommendations

1

Adoption & spread

How quickly AI tools reached the mainstream and which regions lead the way.

2

Productivity & efficiency

When AI actually speeds things up and when it slows developers down.

3

Code quality & security

A 45% vulnerability rate and the rise of technical debt.

4

Impact on teams

How roles, skills and career paths are changing.

5

Law & governance

Copyright, liability and why bans fail.

6

Strategy & outlook

Recommendations for companies and developers.

Download

Download the study for free

Fill in the form and get immediate access to the full study.

1
2

By submitting, you accept our privacy policy.

FAQ

Frequently asked questions about the study

Answers to the key questions about content, data and methodology

What is the aim of this study?

The study examines how AI is changing programming worldwide: fact-based, critical and free of marketing bias. It focuses on adoption, productivity, code quality, security, legal risks and the strategic implications for companies and developers.

What data is the study based on?

The analysis draws on 130+ international sources, including developer surveys (e.g. Stack Overflow, JetBrains, GitHub), market research (e.g. Gartner, Forrester), academic studies (e.g. METR, Microsoft Research) and security analyses (e.g. Veracode, GitClear).

How widespread is AI coding worldwide?

As of late 2025, 84% of all developers worldwide use AI-powered programming tools. AI coding has therefore reached the mainstream and is no longer a niche or experimental topic.

Is a large share of code really written with AI?

Yes. According to the study, around 41% of the world's code is already written with AI assistance. At large tech companies the share lies between 25% and 35%, and it is rising.

Does AI actually make developers more productive?

The answer depends on context: junior developers benefit significantly, while senior developers can even become slower in familiar codebases. Repetitive tasks benefit strongly, complex architecture decisions hardly at all. Blanket productivity promises are therefore not reliable.

Which risks does the study identify?

The main risks are: a 45% security vulnerability rate in AI-generated code, a sharp rise in code duplication and technical debt, declining trust among experienced developers, and unresolved legal questions around copyright and liability.

Why is trust in AI tools falling despite high usage?

The study identifies a trust paradox: while usage is rising sharply, trust in the accuracy of AI is falling. Developers use AI pragmatically, but increasingly critically and with closer scrutiny.

Are AI coding tools a risk to developer jobs?

A third of developers see AI as a potential threat to their jobs. The study concludes, however, that roles are changing rather than disappearing. In future, architecture, review, security and AI skills will be in greater demand.

What role does governance play in companies?

A central one. The study shows that bans often fail (shadow AI). Companies succeed with clear guidelines, training and control mechanisms. Governance is crucial for seizing opportunities and limiting risks.

Who is the study most relevant for?

Software developers (junior to senior), CTOs, tech leads and IT decision-makers, product and engineering managers, as well as investors and strategy leaders.

Ready for AI-assisted development?

Let us develop your AI strategy together. From consulting to implementation, we accompany you along the way.