What is ethics for AI?

Ethics for AI is the value work an organisation does to make sure its AI systems actually hold up against what it says it stands for, not just what the paperwork says. Being legally compliant and being defensible are not the same claim, and most of the difficult calls live in the gap between them: is this a decision we could explain, face to face, to the people it affects, not just to a regulator ticking a box? That is a judgement call, not a checklist item, which is why it needs a named person with the authority and the nerve to make it, not a policy document sitting unread on a shared drive. Governance, compliance, and safety are the raw material that judgement is built from; the value work is turning that raw material into decisions a board, a regulator, and the people affected would all recognise as the same organisation. It is what Virtue & Voltage is built around.

What is the EU AI Act’s Article 4 obligation?

Article 4 requires providers and deployers of AI systems to ensure a sufficient level of AI literacy among their staff. It has been in force since February 2025 and carries no attached benchmark, so organisations must define and evidence competence and capability themselves. A completed e-learning module and a team that can actually reason through a hard AI decision under pressure are not automatically the same thing, so we build that evidence against a credible European yardstick, not just a completion certificate.

What is the free ethics-function maturity self-audit?

Ten statements, most of them testing judgement and culture rather than paperwork, that score whether your AI ethics or governance function has a named owner who has actually exercised their authority, a protected escalation route, and evidence that would survive a board or regulator's question. It runs in your browser, nothing is recorded, and it points you to the right next step. Take it here.

What is the Organisation Toolkit?

A fixed-price pack that establishes and protects the AI ethicist or governance role: charter, interview scorecard, independence policy, reporting line, onboarding framework, and sector tailoring guide, developed independently, from our own casework rather than any single external framework. Sold directly, no engagement required. See what's inside.

What is the AI Ethicist Curriculum?

A licensed curriculum built from the same casework and judgement calls as the toolkit's scorecard, for training and evidencing your own AI ethicist or governance staff: slides, facilitator notes, assessment, and completion records. Licensed directly, single cohort or annual. See what's inside.

Who pays Virtue & Voltage?

Mostly, the organisations we advise. The self-audit is free, and the Organisation Toolkit is sold directly at a fixed price, no engagement required.

Are you independent, or part of a group?

Independent. Virtue & Voltage is a standalone AI Ethics practice. We are not owned by, and do not speak for, any standards body, certification scheme, training provider, or register. Our findings are built from our own casework and judgement.

Do you certify people or issue credentials?

No. We are an AI Ethics practice. We assess competence and capability and design roles and programmes; we do not certify individuals. What we produce instead is evidence, scorecards, reports, sign-offs, built to survive a board or regulator's question on its own merits, not on the strength of a credential behind it.

What AI ethics standards and frameworks does Virtue & Voltage draw on?

We draw on the public international record for competence and role design specifically: EN 18274, the European competence standard for the AI ethicist role, expected to publish in 2026 after passing its final vote, alongside UNESCO's Recommendation on the Ethics of Artificial Intelligence, the OECD AI Principles, and the EU's AI Act and related guidance.Where a board already reports against the NIST AI Risk Management Framework or the ISO/IEC 42001 management-system standard, we read a client's AI ethics function against those too, rather than asking them to adopt a fifth framework.

Who is Role design actually for?

Typically the general counsel or the leadership team, though HR may commission it too. Whoever signs it off, the point of the engagement is the same: give the role real authority and a working escalation path, which is a governance question first and a hiring question second.

Isn’t this just about avoiding risk?

Avoiding risk is the floor, not the ceiling. An organisation that can actually explain an AI decision, to a regulator, a journalist, or the person it affected, keeps trust, and trust is what lets a board move faster on the AI investments that matter, not slower. The field's own direction of travel, from Responsible AI toward what is increasingly called AI stewardship, reflects the same shift: governing intelligence well is closer to a strategic capability now than a compliance line item.

Does the EU Product Liability Directive affect an AI ethics function?

Yes, and it is a different exposure to the one Article 4 creates. Directive (EU) 2024/2853 rewrites the EU's product liability regime so that standalone software and AI systems now count as products, not only physical goods, and Member States must have it in national law by 9 December 2026. It is a strict liability regime: a claimant only has to show the product was defective, that they suffered damage, and that there is a causal link, and courts can presume both the defect and the causal link where a system's technical complexity, the black box problem, makes that hard for the claimant to prove otherwise. In practice, that shifts the evidentiary burden onto the organisation to show its risk assessment, documentation, and update history were adequate, which is the same evidence an AI Ethics Coverage & Exposure Review or an Article 4 readiness assessment already builds, so the two obligations reinforce each other rather than adding separate work.

What if we don’t need you?

We will tell you. An honest "not yet" is worth more to your board than a retainer, and worth more to us than a client who resents the invoice.

What’s the difference between AI ethics, responsible AI, and AI governance?

Three different questions, often flattened into one. Responsible AI asks how to build and deploy a system carefully: testing for bias, documenting limitations, building in safeguards. AI governance asks how to make sure that actually happens in practice: who owns it, who reviews it, who can stop it. AI ethics asks the question neither of the other two can answer on its own: should we be doing this at all, and who benefits and who is harmed if we do? A system can pass every governance check and be built entirely responsibly, and still fail that first question. Virtue & Voltage is built around asking it, not around the other two.

How do you tell a real ethical finding from ethics theatre?

Watch what happens the first time a finding is genuinely inconvenient. A real finding gets a specific answer: yes, no, or "not yet, and here is what would change that." Ethics theatre answers with one of four moves instead: the concern gets called a technical misunderstanding, the person raising it gets called a blocker to progress, someone points at a policy document without testing the finding against it, or the finding gets sent to a committee with no return date. None of those four moves is necessarily dishonest in the moment, that is what makes the pattern so durable. If you keep hearing one of them instead of a real answer, that is diagnostic information about the organisation, not about whether the finding was right.

Does Virtue & Voltage test for human dignity specifically?

Yes, directly, not just as background philosophy. It is the sixth question in our six-question test (who benefits, who is harmed, who decides, who can challenge, at what scale, and whether the person affected would recognise the decision as having treated them as a person, not just processed them), and it is the closing question in the free self-audit: would you be comfortable explaining the reasoning face to face to the specific person it affects, not just to whoever is asking on their behalf.

Independent AI Ethics advisory · live AI risk

Ethics for AI is the missing role.

Virtue & Voltage makes that judgement real: a named role, built against the international standards record, evidenced enough to hold up to a board or a regulator.

Book a conversation See what we do

Voltage is the difference between two poles: your virtue, and what your AI actually does. Ethics for AI is the current that runs across it: the judgement, measured, directed, and signed, that keeps the two poles from drifting apart.

01 · Start here

Before an engagement, two ways to see where you stand.

Not every organisation is ready to buy an engagement. These two are free or fixed-price, need no scoping call, and tell you honestly whether you need one yet.

1. Self-audit (free)→ 2. Organisation Toolkit (fixed price)→ 3. An engagement (only if you need one)

Free · 5 minutes · no email required

The ethics-function maturity self-audit

Ten questions against a recognised competence framework for AI ethics roles, scored in your browser. Nothing you enter is sent anywhere. The result tells you which of the six gaps to close first, and what that costs to fix.

Take the self-audit

Fixed price · no engagement required

The Organisation Toolkit

The charter, the interview scorecard, the independence policy, the reporting line, the onboarding plan, and the sector tailoring guide for an Ethics for AI function, sold as one editable pack. Everything needed to stand the role up and keep it from being decoration.

See what's inside

02 · What we do

Four questions we get asked, and answer.

Ethics for AI is a specific, ongoing job: these are the four engagements a general counsel or a board actually needs to hold it, delivered by someone who signs their name to the finding.

Article 4 readiness

The EU AI Act's AI literacy duty has been in force since February 2025 with no benchmark attached. We are not a training provider bolting on a course: we make the ethics judgement about what “sufficient” actually means for your risk, then build the evidence pack that proves it.

€10k – €30k

Annual · evidence goes stale

AI Ethics Coverage & Exposure Review

A board-level read of which AI duties you carry, which competence your people actually cover, and where you have no cover at all. Ethics reframed as an uncovered exposure.

€5k – €15k

Annual or on change

Role design

An ethicist who reports to the person who owns the launch date has no way to act on what they find. We design the role with the authority, independence, and escalation to actually hold.

€8k – €20k

One-off, then referrals

Expert witness & second opinion

An independent AI ethics opinion for when a court, a regulator, or a board needs one it can rely on: competence, governance, and standards, reasoned and signed by name. Rare, high signal, on a day rate.

Day rate

Case by case

03 · Why now

Article 4 is in force, and nobody has a yardstick.

Why this matters now

Every organisation in scope is improvising its AI ethics.

The AI literacy obligation landed in February 2025 with no attached benchmark, so most organisations are left holding a completion certificate with no way to know if it is enough. We work from real casework and named, checkable judgement, not a certificate that stopped meaning anything the week it was issued.

Feb 2025

Article 4 came into force: the clock on an undefined ethics obligation has been running ever since.

Name attached to every finding: judgement a board can actually hold someone to.

Named accountability is a resourcing choice, not a mystery, and the organisations that make it are still the minority: only 14% of companies have clearly defined at leadership level who is actually accountable for AI governance (Logicalis CIO Report 2026). Closing that gap is what lets the other 78% of business leaders stop guessing whether they could pass an independent AI governance audit within ninety days (Grant Thornton, 2026 AI Impact Survey of 950 senior leaders).

04 · Who we are

A small, senior practice that puts its name on the finding.

Virtue & Voltage is an independent AI Ethics practice: a handful of senior practitioners who have sat on the inside of governance, law, ethics, operations, and AI, across a range of industries, and who tell a board exactly what they see.

Independent

No axe to grind

What we sell, the advice, the free self-audit, the Organisation Toolkit, we sell under our own name, at a fixed price, with nothing behind it.

Senior

The room, not the slide deck

The person who scopes your work is the person who does it and signs it. No pyramid of juniors, no findings diluted through a brand. Judgement you can hold a named individual to.

Plain

We'll tell you no

If you don't need us, we say so. An honest "not yet" is worth more to your board than a retainer, and it is how we would want to be treated.

More about the practice

05 · What clients say

Judgement other people were willing to rely on.

Attributed by role and sector, not by name, the same discretion we extend to every engagement.

"[Placeholder quote -- replace with a real, permissioned client testimonial before enabling this section.]"

[Role][Sector] · placeholder

"[Placeholder quote -- replace with a real, permissioned client testimonial before enabling this section.]"

[Role][Sector] · placeholder

"[Placeholder quote -- replace with a real, permissioned client testimonial before enabling this section.]"

[Role][Sector] · placeholder

This section is switched off by default and not visible on the live site. Set TESTIMONIALS_ENABLED to true near the top of the page script once real quotes replace the placeholders above.

Bring us the question your board hasn't answered yet.

One conversation, no account to create, no sales sequence. We tell you whether we can help, and if we can't, who can.

Book a conversation See the services