7 August 2026

AI Is Reshaping the Technology Landscape for Developers

The Octoverse 2025 Report

GitHub’s Octoverse 2025 report shows the rise of TypeScript, which has overtaken Python and JavaScript, and shell scripts, which have overtaken C++ in the top 10.

Why TypeScript? Because its type system makes the code more resilient to type errors, which accounted for 94% of the JS code errors generated by LLMs (study cited in the report).

(Source: Octoverse 2025 Report)

AI doesn’t make languages better—it makes them more accessible

AI doesn’t make languages more effective or turn bad tools into good ones; it simply makes them accessible to a wider audience.
AI makes it easier to use tools or languages that are perceived as complex or difficult, thereby eliminating the “ease-of-use” bias among developers who, before AI, favored more accessible solutions at the expense of performance.

This aspect of the report seems quite positive, as AI encourages the use of a more powerful tool or language that is better suited to the problem at hand, and strongly influences the choice of technologies adopted by developers (or rather, by their favorite code assistant!).

The somewhat concerning aspect of this report is that 80% of young developers joining GitHub use Copilot within their first week.
AI is now part of the standard toolkit for web developers, just like an IDE; beginners start their learning journey directly with code assistants—hopefully following best practices and using “learning” mode.

An unprecedented amount of code

The volume of code produced has never been greater, proof that code assistants are driving the generation of many more lines of code… which will need to be reviewed, tested, and deployed…. At DATASOLUTION, we integrateAI into web development to support our technological excellence.

Security Under Pressure

According to the report, the use of security checking tools (such as Dependabot, CodeQL, etc.) continues to rise, reflecting growing global awareness of these “supply chain attack” issues.

We recall the Shai-Hulud worm from September–November 2025, which compromised hundreds of NPM packages.

However, these scans show a 172% increase in “Broken Access Control” errors, a significant portion of which is believed to be caused by LLMs.

These security scans are therefore more essential than ever, as AI-generated code can use a library without the developer’s knowledge or inject unsafe code; safeguards must therefore be put in place.

The New Rules of Development in the Age of AI

So now more than ever, we must:

- Pay close attention to coding rules and other guidelines provided to the AI
- Train developers on the proper use of coding assistants
- Maintain human oversight in code reviews and design
- Equipping code-based factories with SCA/SAST tools

Work with small increments of code that are easy to review, validate, and deploy, rather than setting out to build cathedrals… without foundations or guardrails.

Contact our teams to transform your development with AI.