> For the complete documentation index, see [llms.txt](https://vector-privacy.gitbook.io/vector-privacy/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://vector-privacy.gitbook.io/vector-privacy/vector-messenger/more/ai-and-transparency-policy.md). # AI & Transparency Policy {% hint style="success" %} Vector Messenger is not vibe-coded and enforces a Zero Blind Commits Policy. {% endhint %} Example: This page was written by hand and grammatically reviewed by AI. ## Introduction It was brought to our attention that some people were unsure whether Vector was "vibe-coded," during a time when AI is becoming increasingly prevalent and many people are outsourcing their thoughts, skills, and work entirely to AI with little to no human oversight. This is a concerning practice, particularly when there is no professional intervention or review involved. This prompted us to be more transparent about our thoughts and practices, because the public deserves that clarity. The code is open-source, and so should be the vision, values, and workflow behind it. ### What is Vibe Coding? To clearly define what vibe coding is and is not, we need to draw a clear line. Every person likely has a different definition, but most can agree it involves AI in some capacity. The real question is how much. Vibe coding generally refers to people with little to no programming experience using AI to write everything for them with zero review outside of AI itself. In that sense, the code is written and reviewed entirely by AI, with no meaningful human involvement. > **By that definition, Vector Messenger is not vibe-coded and never will be.**
## Zero Blind Commits Policy Vector Privacy is built by a small, independent team with contributors from around the world. Every single pull request and commit on GitHub is manually reviewed by the founder and senior developer. Not a single commit or contribution, regardless of whether AI was involved in writing it, has ever been pushed without review by the senior developer and other team members. AI is not perfect, and neither are human developers, but peer review is a crucial part of quality control and the foundational reason why open-source technology is often more trustworthy and resilient in practice. *** ## Transparency From the very beginning, Vector has been built in public, both by the technical definition of free, open-source software and from a social perspective. Vector's development began around January 2025, at a time when AI tools were not nearly as widespread as they are today, and AI was not used in Vector's development at that stage. Vector has not only included the community from the very start in development, design, and technical discussions, but has actively encouraged even greater engagement and open communication with the public over time. In other words, nothing is kept secret. Everything the core developers are thinking and working toward is communicated openly. The morals, vision, and goals of the project have been defined and shared from the beginning, including what Vector will always aim to do and what it will never do, such as running ads or issuing tokens. With that said, Vector does incorporate AI into its workflow, as do many talented developers, teams, and projects. AI is used to assist with research, code auditing, frontend development, and backend troubleshooting. It is primarily used as an efficiency tool and an additional layer of peer review, as advanced AI models can often process and synthesize information at a scale that complements what a small team can cover alone. That said, it always requires human intervention and oversight, and should be treated as a tool rather than a replacement for professional judgment. {% hint style="warning" %} Vector does not use AI to vibe-code or blindly generate anything related to privacy or security. All privacy and security decisions are written, reviewed, and verified by the core team directly. {% endhint %} *** ## Vector's Workflow ### Stage 1 - Ideation The first stage of development is when new ideas emerge, either from the core team or from the community. Each idea is discussed and analyzed like any problem worth solving: * What is the end goal? * What is the most efficient method to achieve this? * What are the trade-offs or bottlenecks of that method? ### Stage 2 - Design Most ideas require some form of user interaction, so the UX/UI and user flow are planned out in [Figma](https://www.figma.com/) by the lead designer before any code is written. This ensures the experience is intentional and well thought out before development begins.
### Stage 3 - Development After everything is visually mapped out, developers work to implement the design in code. In most cases, this means running a local production build from the latest branch on GitHub and testing changes in real time before committing, submitting a pull request, and merging into the main codebase. ### Stage 4 - Auditing In addition to the team and contributors compiling their own builds to test new features and identify bugs, issues, and discrepancies, Vector utilizes AI to help run a wide range of scenarios in a short amount of time, hardening the code and making it as reliable as possible. This is not done through AI alone, but as an additional layer within the Vector workflow. AI should never be the sole source of validation. Human oversight is always required, and no AI result should be taken at face value without verification by the team and its peers. *** ## AI Usage In summary, AI is utilized as an additional member of the workflow rather than a replacement for any part of it. It is one more tool added to optimize the process and provide a level of coverage and attention to detail that would be difficult for any small team to match alone. That said, it is experimental, far from perfect, and should never be left unguided or unverified. Always cross-reference what AI produces with your own judgment and that of your peers. Trust, but verify.
Fields
Research & Development
Auditing
Frontend Development
Backend Troubleshooting & Optimization
*** ## Disclaimer Vector Messenger was not vibe-coded, nor will it ever be. It was planned and developed before the era of large-scale AI adoption, and the team has been working in their respective fields long before AI became a mainstream reality. As a result, AI is used as one more tool in the workflow, not the foundation on which everything depends. As outlined above, good practice with AI means never trusting its output entirely without cross-referencing and verifying it yourself. Use it as a tool, but never depend on it.