Skip to content
AIAC AI ASSURANCE COUNCIL

Contributor guidelines

We would rather publish a sharp argument we disagree with

The Council's blog carries practitioner writing on AI assurance, security, governance, and regulation — people describing what they have actually seen, and what they think it means. Not press releases, and not summaries of things already said elsewhere.

Reference
AIAC/EDI/2026/01
Issue
1.0
Effective
8 September 2026
Next review
September 2027
Status
Contributions are by invitation only

§ 1 — Status

By invitation, for now

The Council is not open to unsolicited submissions. If you have received an invitation, these guidelines tell you what we are after and how the process runs. If you have not, there is nothing to send yet — and a pitch to the general inbox will not reach an editor.

The audience is engineers, risk and compliance leaders, auditors, legal and finance professionals, and policy people. Practitioners, in short, who will be asked to defend a judgement and who can tell substance from hype at a glance. Write for them.

§ 2 — What we publish

Your own take, argued

A clear argument, a lesson from real work, or a contrarian read on something the field is currently nodding along to. The perspective is the whole point. A safe survey of the consensus is the one thing we will not run, however well written — it is already available everywhere else.

These are the themes we are commissioning against. An idea outside them is welcome if it is genuinely for this audience; raise it in your invitation thread before you write it.

AI assurance
Testing, evaluation, evidence, and what “assured” should actually mean.
AI security
Model, agent, and pipeline security; red-teaming; failure modes seen in practice.
AI governance
Frameworks as they are lived — ISO/IEC 42001, the NIST AI RMF — and where implementation is theatre.
Agentic AI
Autonomy, identity, control, and the assurance problems that follow from them.
AI in financial services
Deployment under supervisory constraint: MAS expectations, DORA, and beyond.
AI in legal practice
Real workflows, liability, and professional-responsibility questions.
AI regulation and policy
The EU AI Act, national frameworks, and the places policy meets operational reality.
Certification and workforce
Skills, credentials, hiring, and how capability is actually built.

§ 3 — Format

Six hundred to nine hundred words

Long enough to make a point and short enough to respect the reader. One argument, led with rather than buried — if your thesis first appears in the eighth paragraph, the first seven are the ones to cut.

Send a working title and a one-line summary we can use as the excerpt. Write in plain, direct prose: a practitioner talking to peers, not a whitepaper. Cut the throat-clearing opener; nobody has ever finished a piece because it began with “in today's fast-moving landscape”.

§ 4 — Authorship

The thinking has to be yours

This is the line we hold hardest, because a body that certifies competence cannot publish machine-written opinion under a practitioner's name and expect either to be believed.

The article must be substantially your own work.The argument, the judgement, and the voice have to be yours. AI tools may help you draft, tighten, or research, and we have no objection to any of that — but you are accountable for every claim in the piece, and something essentially generated will not be published.

Disclose material AI assistance. A single line at the end is enough: “Drafted with AI assistance; the views, claims, and errors are my own.” Disclosure is never held against a piece. The absence of it, where it was warranted, is.

AI-generated images are welcome for the header or for illustration, with no disclosure needed. The exception is anything that carries information: never generate a chart, a figure, a dataset, or a quotation.

It must be unpublished and yours to give. No plagiarism, and nothing already live elsewhere. A reworked version of your own earlier writing is fine if you tell us where it ran.

§ 5 — Sources

Cite the instrument, not a summary of it

Link to primary sources wherever they exist — the standard, the regulation, the paper, the repository, the company's own page — rather than to somebody's account of them. It is the same discipline the framework guides on this site are written to, and it is the difference between a claim a reader can check and one they have to take on trust.

Back anything non-obvious. Where the evidence is thin or the position is contested, say so in the sentence rather than in a footnote — a hedged claim with its uncertainty stated is worth more here than a confident one that turns out to have been a guess. Two to six links is a healthy range at this length.

§ 6 — Limits

What we will not run

Vendor pitches. Mention your product once, in context, where it genuinely bears on the argument. A piece whose shape is an advertisement gets declined, and nothing published here may suggest that the Council has assessed, approved, or accredited a product or service.

Unsupported certainty. If a claim is your opinion or your bet, frame it as one.

Filler. If a paragraph could be deleted without loss, delete it before we do.

Anything defamatory or confidential, or that you do not hold the rights to publish. Assume your employer's counsel will read it.

§ 7 — Independence

A byline is not an endorsement

Contributors are often candidates, members, or people working toward a Council credential, so this needs saying plainly. Publishing here is not an endorsement of you, your employer, or your product, and it confers no advantage of any kind in an examination, an application, or a certification decision. Assessment is separate from editorial, and stays that way.

The reverse holds too: we will publish an argument that disagrees with the Council's own position, and have no interest in a blog that only agrees with us. What we will not publish is a piece that implies the Council has certified something it has not — the boundary is set by the terms of use and the Code of Professional Conduct, which bind credential holders whether they are writing or not.

§ 8 — Attribution

Every piece carries your byline

Send us, with the draft:

  • Your name and title.
  • A bio of one or two lines.
  • One link — LinkedIn, your company, or a personal site.
  • A headshot, if you would like one shown.

If you hold a Council credential, we will show the post-nominal alongside your name. We will not add one you have not earned, and we will not remove one you have.

§ 9 — Process

How it works

  1. One

    Reply to your invitation

    With a full draft, or with a paragraph on your angle if you would rather confirm the direction before writing.

  2. Two

    Editorial review

    We read for fit, clarity, and accuracy. We may come back with notes, or ask for a source we could not find ourselves.

  3. Three

    Light editing

    We copy-edit, and may adjust the title or the formatting. We do not change your argument without asking you first.

  4. Four

    Publication

    Once you approve the edit, we schedule it and tell you when it is live.

An invitation is not a guarantee of publication. We may hold a piece because the angle is not right, or because the timing is not — and we will always tell you which, and why.

§ 10 — Rights

You keep your work

You retain ownership and copyright. By sending a draft you grant the Council a non-exclusive right to publish it here and to promote it on the Council's channels, always with attribution to you.

You are free to republish elsewhere afterwards; a link back is appreciated but not required. If you ask us to take a piece down, we will, and we will say on the page that it was withdrawn at the author's request rather than leaving a silent gap.

§ 11 — Submitting

Been invited?

Reply to your invitation thread with the draft, or with your angle if you would rather agree the direction first. That thread is the fastest route — it already has an editor on it.

Anything else — a question about these guidelines, or a note that you will be late — reaches an editor at editor@aiassurancecouncil.org. That address is editorial only. It is not a route around the invitation: an unsolicited draft sent there will get a courteous reply and nothing more.