The recent rapid advancement of AI/LLM technology has introduced considerable confusion, disorientation, and with it alarming amounts of false promises, snake oil sales tactics, and sleazy shilling. Our duty as software professionals is to prevent such degradation of quality by insisting on a high watermark professional code of ethics.
ACG Manifesto
Embracing Innovation
We, software development professionals, are exploring the innovations made available by the development in the Artificial Intelligence space.
Demanding Quality
We are concerned about the proliferation of half-baked, barely verified and tested content that current AI/LLM technology is notorious for delivering.
Ensuring Safety & Security
We insist that a proper harness be provided to ensure engineering quality, safety, security, and legal liability for all products delivered with the assistance of AI technology.
Protecting Humans
We advocate the need for psychological safety of humans who are exposed to any products and content generated by AI.
Read Full Manifesto
Explore the complete code of conduct including our principles on user agency, informed consent, and the professional duty to refuse unethical work.
The ACG Manifesto · Four Pillars
What the Guild actually stands for
Ethics you can fail a build on. Four engineering positions we hold when a model is in the loop, written for people who ship production systems.
Adopt AI deliberately
We put LLMs in the toolchain when they measurably raise output quality, not because a vendor deck said to. Curiosity, evaluated.
Reject unread output
No merged PR nobody read. No untested generation. No plausible-looking hallucination shipped to prod because the diff was long.
Harness the model
AI-assisted delivery ships with the same tests, reviews, security scans, and liability chain as human-written code. No exceptions for speed.
Disclose to the human
Users get to know when a model is on the other side, what it was trained to do, and a clear path out when its output harms them.
The Guild also works.
Independent, vendor-neutral assessment of AI-assisted codebases (a ladder from a first-look diagnostic to continuous security assurance). We sell no tooling, so every recommendation is the one we would give a friend. Start with the self-assessment — it ends with an honest answer, not a sales pitch.
Sign the Manifesto
Join fellow professionals committed to excellence in AI-assisted development
Signatories
Join the Movement
Stand with us in championing integrity, transparency, and professional excellence in AI-assisted development.