STRATEGIC LEADERSHIP IN HUMAN RIGHTS AND HUMANITARIAN BUSINESS PRACTICES · 2026 EDITION

Course Revision Companion

Last updated on 11 August 2026. Everything on this page comes from the course as it was actually taught — the slides we used in class and the course's own reading notes, nothing imported from elsewhere.

How to use this document. This page was made to accompany you while you revise, and it is organised around a simple idea: revision works better when it starts from who you are and what you have already chosen to care about. There are three parts, and you are not expected to read all of them in one sitting. The first part starts from you — first from the profile quiz you filled in on the very first day of the course, and then from the case territories that emerged from the portfolio topics this class actually chose. The second part walks through the whole course, one day at a time, with a short self-test at the end of each day so you can check your understanding as you go. The third part is an index of the ideas that kept returning across the days, with a note on where each one lives. A suggestion, offered honestly: read your own territory first, because that is where your energy already is. And when you have done that, read one territory that is not yours. It is not extra work — it is precisely the kind of crossing that shows an examiner the difference between understanding a course and having memorised one corner of it.
The tools you played with, all in one place. Over the thirteen days you didn't only listen — you answered, negotiated, simulated and voted. All of those tools are still online, and revisiting them is one of the most pleasant ways to revise, because each one holds a memory of how a concept felt in practice.

Part 1 · Revision material organised around you

This first part offers you revision material in two complementary ways: starting from the profile you received on the first day, and starting from the case you chose for your portfolio. Both routes lead into the same course — they simply enter it from different doors, and you are welcome to use whichever feels more natural, or both.

1a · The quiz you answered on the first day

On the first day of the course, you filled in a short quiz: fourteen forced choices, none of them comfortable, which gave you a profile. You can retake it here (humanitsim.pages.dev/compass) whenever you like, or simply retrieve the result you remember from that first morning. This section is dedicated to giving you revision material connected to each of those profiles. Two things are worth saying before you read yours. The first is that the profile was never a box — it was a snapshot of your inclinations on one particular day, and a course of this kind is designed to move people, so it is entirely possible that you would answer differently today. If you retake it and the result has changed, that change is itself worth five minutes of reflection: read both routes, and ask yourself what the course did to you. The second is that every profile below comes with a stretch — the part of the course your inclination tends to walk past. It is described gently, but please take it seriously: it is usually where the most valuable revision hides.

If you like, choose your profile here, and the page will light up your route. Your profile card, the days that belong to your path, and the case territories that fit you will be clearly highlighted all the way down the page — the rest stays visible, just quieter. Your stretch days are marked in a warmer colour, as a friendly reminder. You can change your choice, or clear it, at any time.
The Advocate
When you answered the quiz, you leaned toward voice — toward the conviction that when a person's rights are being violated, someone has to say so, clearly and out loud, even when it is inconvenient.
If that is you, Days 1 and 2 are your foundations, and they deserve a slow read: they explain where rights actually come from, and they carry Arendt's difficult question about who gets to claim rights at all — a question every advocate eventually meets in practice. Day 5 will give you a sentence you are likely to use for the rest of your working life: that business in a conflict zone is never neutral. And Day 10 is, in a real sense, your day — it is entirely about who is heard, who is allowed to challenge a decision, and what actually changes because of the challenge.
Where to stretch. The part of the course your inclination tends to walk past is Day 3, the day about numbers. Advocacy that cannot engage with measurement gets dismissed in every boardroom it enters — not because it is wrong, but because it can be waved away. Learn the indicators well enough to argue with them, and your voice becomes much harder to dismiss.
When you reach the case territories below, begin with business in conflict, labour rights, and access and development — that is where advocates' portfolios tend to live.
The Steward
Your answers leaned toward care — toward protecting people and resources, and keeping an organisation running without causing harm along the way.
Days 5 and 6 will feel like home to you. They treat heightened due diligence and the risk register not as paperwork but as an operating system — which is exactly how a steward already thinks. Day 9 then shows you what stewardship looks like when it reaches board level, with real duties and real records, and Day 12 asks what careful stewardship becomes when the old canon meets very new technology.
Where to stretch. The day you might be tempted to treat as someone else's is Day 8, and it would be a real loss. Stewardship depends on structures, and Day 8 stages the course's central collision — Wettstein's argument that companies have positive duties, against Choudhury's demonstration that corporate law quietly installs the opposite default. It explains, without blaming anyone, why good stewards so often fail inside badly built companies.
In the territories below, start with environment and disaster, and with supply-chain due diligence.
The Strategist
You answered for the long game — for positions and leverage, for understanding what the other side of the table wants before deciding what to offer it.
Day 4 will reward you first: it teaches forum routing, the genuinely strategic question of which defendant, in which forum, through which legal route — and why cases live or die on that choice. Day 7's simulation and Day 10's Nordhavn negotiation are your natural habitat, and you will probably enjoy them; let yourself also be examined by them, because both are designed to reveal what a strategy costs the people who are not at the table. Day 9 shows the governance machinery any strategy has to survive.
Where to stretch. Your stretch is Day 11. Rules that outlast cleverness are being written for artificial intelligence right now, and a strategist who ignores the rule-writing table is planning brilliantly for the wrong decade.
In the territories below, begin with business in conflict and with leadership and governance.
The Operator
Your answers leaned toward delivery — the conviction that plans are only as good as what actually happens, and that someone has to make it happen.
Day 6 is your day, and it says so almost explicitly: due diligence there is treated as an operating system rather than a policy document, which is how you already see the world. Day 7 introduces deprivation cost — the human price of relief arriving late — which turns delivery speed, your speciality, into an ethical quantity. And Day 11's deployment exercise is operations under the hardest constraints the course can build.
Where to stretch. Your stretch is the beginning of the course, Days 1 and 2. The rules you operate under have origins and reasons, and it is tempting to skip them in favour of the practical days. Please don't: knowing why a rule exists is what separates an operator from a bureaucrat, and it changes how you improvise when the rulebook runs out.
In the territories below, start with environment and disaster, and with supply-chain due diligence.
The Investigator
You answered for facts — for finding out what actually happened before anyone starts arguing about what it means.
Day 3 belongs to you almost entirely: whether we can trust the numbers, the difference between detecting more cases and there being more cases, and the uncomfortable finding that the big ESG ratings agree with each other at a correlation of about 0.3. Day 4 shows you which kinds of evidence survive a courtroom, and Day 6 gives you the risk register's most honest column — the evidence gap.
Where to stretch. Your stretch is Day 10. Evidence that never reaches an audience changes nothing, however good it is. Verification needs a route into real decisions, and that route is engagement — the part of the work investigators most often leave to others.
In the territories below, begin with supply-chain due diligence and with forced labour.
The Communicator
Your answers leaned toward the story — toward the understanding that how something is told largely decides what people end up doing about it.
Day 7 will be your critical mirror, and it may sting a little: the campaign laboratory and the Brand Aid literature examine the industry that monetises exactly your talent, and they do it with respect but without mercy. Day 8 is about the distance between projection and practice — between what a leader communicates and what their institution does. Day 10 shows you disclosure, which is what telling looks like when it becomes an institutional obligation.
Where to stretch. Your stretch is Day 3. Behind every honest message there has to be a mechanism and a metric — something that actually works, measured in a way that could fail. Learn to check before you tell, and your communication gains the one quality no technique can fake, which is being right.
In the territories below, start with access and development, and with the washing chapters of AI and digital rights.
The Appraiser
You answered for judgment — for the willingness to weigh something, score it, and say plainly whether it worked, knowing that someone has to.
Day 3's indicator discipline is your home ground, and Day 6's risk register is your instrument. Day 10 gives you what might be your creed: the verification questions — who checks, how often, who sees the results. And Day 11 hands you the frontier, with a detail you will appreciate: in the AI simulation, the function that quietly went missing was 'Measure'. The course is telling you that your kind is needed and scarce.
Where to stretch. Your stretch, perhaps surprisingly, is the very beginning — Days 1 and 2. Every measurement serves a purpose that someone chose, and the foundations explain what the numbers are ultimately for, and what they will never capture. An appraiser who has read Arendt weighs differently.
In the territories below, begin with supply-chain due diligence and with AI and digital rights.

1b · Revision material organised around your portfolio case

These eight territories come from the portfolio cases this class actually chose — nothing abstract, just the ground the cohort is standing on. Find yours, read its shelf, and try its questions out loud. Then follow one borrowing into someone else's territory: that's not extra work, it's how examiners tell depth from memorization.

Business in conflict & atrocity complicity
Home ground if your portfolio lives near: Unocal · Kiobel · Lafarge Syria · Meta/Myanmar · Telenor Myanmar · arms exports
Day 4Day 5Day 6Day 9Day 11
Readings worth returning to first — what each one argues, and how they speak to each other
Kolieb (2020) — Human-rights law alone isn't enough in conflict. IHL is non-derogable, extraterritorial, and binds persons directly — it closes the liability gap Day 4 leaves open.
Joseph (2004) with Dodge (2019) — Unocal made corporations ATS defendants through knowing practical assistance; then the Supreme Court contracted the ATS by avoidance. Together they explain why your case's forum kept moving.
UNDP (2022) with Traoré & Cirimwami (2024) — Diligence doubles in conflict: impacts on people AND on the conflict itself — business is 'never neutral'. And when conduct diligence reaches its limit, the honest options are suspension or disengagement. Read them as one argument: method, then exit.
Kobrin (2009) — Talisman is your ancestor case: private political authority without accountability machinery, and the exit-relocates-harm question every board in your territory eventually faces.
Olsen & Bernal-Bermúdez (2023/24) — Over 30% of Latin American abuse claims are state-enabled. Compliance systems cannot monitor the state that hosts you — the coefficient behind 'the conflict won't ignore you'.
Garon (2022) — From Zyklon-B to algorithmic amplification, the knowledge question persists. If your case touches platforms, this is your bridge into Day 11.
Questions you should be able to answer about your own case
  1. Route your case through the three gates (defendant · forum · route): which gate was decisive, and was it ever actually decided?
  2. What would heightened due diligence have had to find, and by when, to change the company's course — and would Traoré & Cirimwami's limit have required exit?
  3. Olsen & Bernal-Bermúdez say your perpetrator may be the state itself. What does that do to your company's compliance architecture?
Worth borrowing from the supply-chain disclosure territory
Borrow OECD (2016): the five-step minerals system is your due-diligence chapter 'with plumbing' — trace, respond, audit the chokepoint, report.
Worth borrowing from the ai and digital rights territory
Borrow ICRC (2019): the turtle-classified-as-rifle problem is your checkpoint, automated. Your knowledge standard is about to be industrialised.
AI, surveillance & digital rights
Home ground if your portfolio lives near: Workday/Mobley · Project Maven · GenAI harms · Amazon surveillance · Safe-City pilots
Day 3Day 8Day 10Day 11Day 12
Readings worth returning to first — what each one argues, and how they speak to each other
Fioravante (2024) — Digital washing: a perception-based business case can be satisfied by managing perception. The five detector questions — including 'who owns the ethics debate?' — are your first analytical move.
Jones (2023) — Human rights beat 84+ AI ethics codes on maturity, consensus and machinery. The cause of the mess is the empty chair — the same structure-installs-the-default lesson as Day 8.
EU AI Act (2024) with Hogan & Lasek-Markey (2024) — The binding risk pyramid names your stack high-risk — but it's product-safety law with a military carve-out; rights sit in the purposes, not the machinery. Know both the promise and the ceiling.
NIST AI RMF (2023) — GOVERN–MAP–MEASURE–MANAGE, voluntary, no substantive floors. In the Day 11 simulation, Measure is what went missing — the softest joint is your exam material.
Beduschi (2022) — Anticipatory humanitarian AI, priced by data quality, bias and consent. Bias-as-denial is Arendt's claimability collapse, automated.
Henriksen (2024) — Big Tech humanitarianism runs care and control from one sector; returns are reputation, data, talent, markets. The complication your optimistic sources need.
Questions you should be able to answer about your own case
  1. Take one real responsible-AI statement from your case and run Fioravante's five questions plus the course's sixth: who verifies, at what interval, with what publication rule?
  2. Where exactly does your technology sit in the EU AI Act's pyramid — and does the carve-out trilogy (treaty scope, CSDDD, Art. 2(3)) touch it?
  3. Design the clause set your deployment needs to survive crisis: scope, disclosure, verification, remedy — each with actor, verifier, consequence.
Worth borrowing from the foundations of the course (Days 1–2)
Borrow Arendt (1951): 'the right to have rights' is what your biometric registration case is really about — membership before rights, now decided by a classifier.
Worth borrowing from the forced labour territory
Borrow LeBaron (2021) via WHO (2024): the underpaid annotators behind your model are the garment supply chain rediscovered. Value-chain due diligence transplants directly.
Forced labour, trafficking & migrant work
Home ground if your portfolio lives near: Finnish wild berries · Qatar 2022 · Ukrainian refugees in Slovakia
Day 1Day 2Day 3Day 6Day 9
Readings worth returning to first — what each one argues, and how they speak to each other
LeBaron (2021) — Sub-poverty wages plus debt are forced labour's recruiting mechanism — and audits ignore wages and purchasing practices. The buyer manufactures the risk its own audits then fail to find.
ILO (2019) with Maul (2019) — The Universal Labour Guarantee re-arms 1919's living-wage floor for all workers regardless of contract; tripartism explains how such norms travel — and where capture creeps in.
Tyc (2021) — Thirty years of the unadopted social clause: trade access conditional on labour standards, stuck between exigibility and protectionism. Your case's enforcement gap has a history.
Choudhury (2023) — Asset partitioning plus shareholder purpose install the instrumental default by law — Rana Plaza is the emblem. Explains why your supply chain behaves as it does even under mandatory HRDD.
McCorquodale & Neely (2022) — The directors' oversight duty maps 'almost exactly' onto UNGP 17 due diligence — Antuzis is where mistreating workers became the director's personal problem.
Moyn (2018) — He exempts labour rights from his floors-critique: countervailing power, not charity. Your territory is the one he thinks still works.
Questions you should be able to answer about your own case
  1. Apply three ILO forced-labour indicators to your case's established facts: what does each prove, and what can it not prove?
  2. A company in your sector doubles its identified cases. Using Day 3's detection-versus-incidence distinction, argue the good-news and bad-news readings.
  3. Where does LeBaron's audit blind spot sit in your case — and what would an audit have had to inspect to catch what courts or journalists later found?
Worth borrowing from the supply-chain disclosure territory
Borrow Dobbin & Kalev's surgeon test: outcomes per group, not initiative counts — the audit rule for every 'S' metric your case's company publishes.
Worth borrowing from the conflict and complicity territory
Borrow Lubanga and Joseph: where forced labour meets conflict, the individual-liability route stays open after the corporate forum closes.
Labour rights & freedom of association
Home ground if your portfolio lives near: Samsung Chennai · Beed district workers
Day 1Day 2Day 6Day 9Day 10
Readings worth returning to first — what each one argues, and how they speak to each other
Maul (2019) — Tripartism regulated worker voice into the legislature itself. FoA is the enabling right because it builds the counter-power all other standards lean on.
ILO (2019) — Time sovereignty, human-in-command, the re-armed wage floor — the forward agenda your case's workers are effectively litigating in the present.
McBarnet (2007) — Responsibility runs beyond law, through law, and for law. Antuzis is beyond-law pressure becoming doctrine — the path your case may be on.
Dobbin & Kalev (2022) — Culprit-correcting programmes backfire; system redesign with an owned number works. The test for every HR initiative your company waves at the union.
Ford, Gillan & Thein (2024) — The NCP system: no teeth, yet 40%+ agreements — and 70% of complaints filed by Northern CSOs for Southern communities. Represented is not present; your workers know the difference.
Questions you should be able to answer about your own case
  1. Why is freedom of association called an enabling right — what concretely does it enable in your case?
  2. Map your company's response onto McBarnet's grid: which moves ran beyond law, which through it, and which were creative compliance?
  3. At which engagement level (consultation / participation / steering) did the company meet its workers — and what level did the risk deserve?
Worth borrowing from the leadership and governance territory
Borrow Choudhury (2023): before blaming the manager, check what the corporate structure makes rational. Your union-avoidance story may be a purpose-default story.
Worth borrowing from the environment and disaster territory
Borrow the Vestland stress test: stay, pause, re-consult, exit — the same board grammar applies when the pressure is a strike, not a spill.
Environment, health & disaster
Home ground if your portfolio lives near: Brumadinho · Braskem Maceió · Tropodo · EACOP
Day 4Day 6Day 9Day 10
Readings worth returning to first — what each one argues, and how they speak to each other
Bradshaw & Ojeda on Vedanta, with Wang (2024) — Parent duty via ordinary negligence — the company's own sustainability report as Exhibit A. Wang then flips the perverse publish-less incentive: documented diligence as defence, not confession.
Boyle (2012) with Boyd (2011) and Heinämäki (2022) — Two routes to the environmental right — judges greening existing rights, constitutions writing new ones — and the sobering arithmetic: complied-with procedure still points to ~2.6°C.
Liu & Hoskin (2023) — Fukushima: a state-sized defendant defeats the whole corporate toolbox, and the claim migrates into inter-state law. Test whether your case's real defendant is reachable at all.
Fan, Shao, Wang & Liang (2024) — Deprivation cost — the human price of relief arriving late — turns disaster response into contract design. Your remedy chapter has an operations research literature.
GRI (2016) — Stakeholders include the voiceless via proxies; engagement becomes accounting. Publish the list, the exclusion basis, the topics, the responses — your consultation critique, formalized.
Questions you should be able to answer about your own case
  1. Run the Vedanta logic on your case: what did the parent publish, and what duty might it found — under Wang's regime, does the same document defend them?
  2. Build the five-line risk register the company needed the year before the harm: risk, source, status, involvement, action, evidence gap.
  3. Who in your case could actually have paused the project — and under which clause, verified by whom?
Worth borrowing from the supply-chain disclosure territory
Borrow Nielsen & Villadsen's test — does the language commit to anything measurable? — and apply it to your company's last sustainability report; Vedanta shows where that answer ends up.
Worth borrowing from the ai and digital rights territory
Borrow Beduschi and Pantiris: anticipatory systems, human judgment in the chaotic 72 hours. Your early-warning chapter is being automated as you write.
Supply-chain due diligence & disclosure
Home ground if your portfolio lives near: Tony's Chocolonely · H&M · Bestseller · Kahawa Bora/DRC
Day 3Day 5Day 6Day 9Day 10
Readings worth returning to first — what each one argues, and how they speak to each other
Liang & Renneboog (2020) with UNGC/GRI (2018) — ESG ratings correlate at ~0.3 while capital allocates on them — and the standard-setters concede the measurement gap from inside. Your numbers chapter starts here.
Eccles & Serafeim (2013) with Dobbin & Kalev (2022) — The trade-off is real; only innovation on material issues shifts the frontier — and initiative counts prove nothing. Mechanism plus metric, or it didn't happen.
OECD (2016) and OECD (2023) — Five steps with plumbing — trace, respond, audit the chokepoint, report — and the 2023 addition your territory forgets: responsible disengagement, whose own harm must be weighed.
Martínez San Millán (2021) — Four conflict-minerals instruments audited: premiums and traceability gained; laundering, diversion and a de facto embargo alongside. Disclosure never prohibits — the finding your policy chapter must survive.
Smit et al. (2020) with Principale (2023) — As HRDD hardens, 'reasonable' becomes a standard of care filled by actual practice — and the CSDDD fuses it into board duty: do, document, publish.
Bartels & Schramade (2024) — Capital is the strongest lever and it's starved of comparable data. Reporting into a vacuum is performance — the investor-side twin of the 0.3 problem.
Questions you should be able to answer about your own case
  1. For one initiative in your case: name its mechanism, its metric, and what evidence would falsify it.
  2. Apply Martínez San Millán's audit to your case's favourite instrument: what did it actually change, and what did it displace?
  3. Which of the company's voluntary commitments does the CSDDD harden into board duty — and which remain governance practice?
Worth borrowing from the forced labour territory
Borrow LeBaron (2021): if your audits don't look at wages and purchasing practices, they audit around the mechanism that recruits forced labour.
Worth borrowing from the leadership and governance territory
Borrow Choudhury (2023): even mandatory HRDD risks window-dressing while corporate purpose points the other way. Your compliance chapter needs a structure paragraph.
Access, medicines & development
Home ground if your portfolio lives near: Novo Nordisk patents · coffee value chains
Day 2Day 3Day 10Day 12
Readings worth returning to first — what each one argues, and how they speak to each other
Moyn (2018) — Croesus: perfect compliance compatible with absolute inequality. Rights are floors machinery — your case lives exactly where the ceiling question begins.
Fassin (2012) — Humanitarian government: compassion as a one-way gift that fixes hierarchy. The frame for every access-programme your company publicizes.
Richey & Ponte (2021) with Atal & Richey (2021) — Brand Aid is aid to brands; whoever defines the crisis defines what counts as help — always via consumption. Your CSR chapter's genealogy, from AmEx 1983 onward.
Wettstein (2012) — Positive obligations grounded in corporate purpose — the strongest philosophical case FOR your company doing more than compliance. Then let Choudhury answer him.
Scholz et al. (2024) — COVAX: contract, strategy and reputation — not virtue — drove vaccine firms' engagement; responsible at country level, not globally. The empirical cold shower for both sides.
Elgar Handbook — 'Maximum available resources' is still unmeasurable — the 1966 split built social rights with weaker claimability by design. Your access claim inherits that architecture.
Questions you should be able to answer about your own case
  1. State the strongest version of the business-model defence in your case, then answer it with Wettstein — and answer Wettstein with Choudhury.
  2. Apply Scholz's finding to your case: which lever (contract, strategy, reputation) could actually move the company, and who holds it?
  3. Is your case a floors problem or a ceilings problem, in Moyn's terms — and what follows for the remedy you propose?
Worth borrowing from the foundations of the course (Days 1–2)
Borrow Rorty (1993): sad stories, not arguments, expand sympathy — then watch Day 7's campaigns manufacture exactly that, and decide where legitimate persuasion ends.
Worth borrowing from the leadership and governance territory
Borrow Strand (2024): industrial-foundation ownership makes long-termism rational. Your company's generosity question may be an ownership question.
Leadership, governance & everyone's spine
Home ground if your portfolio lives near: all portfolios — the layer no case escapes
Day 7Day 8Day 9Day 12
Readings worth returning to first — what each one argues, and how they speak to each other
Maak, Pless & Voegtlin (2016) — Value orientation produces style produces political engagement — moderated by governance. The chain beneath every leader your case praises or blames.
Wettstein (2012) against Choudhury (2023) — The course's central collision: purpose-grounded positive duties versus the instrumental default corporate law installs. Agency and structure, one argument from two ends.
Strand (2024) — The Nordic 'critic as resource' reflex explained by ownership structure — purpose reoriented by foundations, not statutes. The counter-experiment to Choudhury.
Scholz et al. (2024) — CEO conviction is real only at the margin, inside structure. Cite it whenever a leadership narrative is doing too much work.
McCorquodale & Neely (2022) — Oversight duty ≈ UNGP 17 — but the duty is owed to the company, with victims outside. The doctrinal engine, and its limit, in one paper.
Reisman (2012) with Ruggie (2017) — Normativity is authority times control — a boycott can outbind a statute; the UNGPs run on three governance systems plus a social norm. Why your 'soft' instruments keep having hard effects.
Questions you should be able to answer about your own case
  1. Which archetype did your case's leadership perform — Steward, Humanitarian, Strategist — and what did the performance cost?
  2. Take one leadership claim from your case: which legitimacy base carries it, and does practice match projection?
  3. Sketch the governance architecture that would have caught your case's failure early: ownership, escalation, record — and who was structurally unable to see it.
Worth borrowing wherever your own case lives
This spine runs through every other profile: wherever your reading list feels thin, the Day 8–9 pairing (Maak → Choudhury → McCorquodale & Neely) supplies the argument's skeleton.
Worth borrowing from the next frontier (Days 11–12)
Borrow Jones (2023): the empty chair is a governance diagnosis. Whatever your case, ask who is structurally absent from the table that decides.

Part 2 · The course, day by day

What follows is the whole course in sequence, one day at a time. For each day you will find the central ideas, the legal instruments and cases that were actually used in class, the exercise we did together, and three questions you can use to test yourself. The days were designed to build on each other, so if a concept feels loose, the day before it usually holds the missing piece.

Day 00 — Course Opening: the course, the programme, and how to use them
The ideas at the centre of the day
· Thirteen days organised in five thematic blocks· Course calibrates continuously to the group's interests and gaps· Executive-education design: minimal shared exigencies, real personal autonomy· Leadership taught as multidisciplinarity, competency-based decision-making, ethical responsibility· "Strategic" means incorporating multi-level, cross-setting interaction into analysis· Capacities trained: structure fuzzy information; reverse-engineer problem structure· Analytic analogy maps structure to structure and declares its limits· Outcome aim: autonomous sensemaking, problem recognition, problem solving
The instruments and cases we worked with
No legal instruments taught; framing references only: Rorty (violence as inhumane treatment); International Law fragmentation described as "the Hydra" (rejected as random metaphor) · Games typology as demonstration analogy: competitive/cooperative, symmetric/asymmetric (chess, rescue operation, regulator vs firm)
What we did together The Role Compass: 14 forced choices, 7 roles, ipsative self-snapshot, retakable any day (live web tool)
Three questions to test yourself
  1. What distinguishes an analytic analogy from a random metaphor?
  2. Name the three components of leadership as the course defines its learning objectives.
  3. What does "strategic" mean in the course title, as opposed to operational?
→ Hands over directly to the Day 1 content deck — the course itself begins.
Day 01 — Introduction to International Legal Aspects of Human Rights & Humanitarian Protection
The ideas at the centre of the day
· Read rights from their end (telos), not their origin· Rights end by expulsion: Arendt's "right to have rights"· Claimability: a right needs addressee, forum, remedy· Rights collapse with war, the guarantor state, or life itself· A right = legal norm plus a holder's enforceable claim· Governance gap: multinationals escaped the state-addressed rights grammar· Justification chosen prices the demand placed on the firm· UNGPs won unanimity by bracketing foundations and inequality
The instruments and cases we worked with
Instruments: ILO Constitution 1919; UN Charter 1945; UDHR 1948; Geneva Conventions 1949; two Covenants 1966; ICCPR Art. 4 (derogation + non-derogable core); CEDAW reservations (Egypt, Iraq, Saudi Arabia); draft UN Norms 2003; Global Compact 1999; UNGPs 2011 (Protect–Respect–Remedy; HRDD: assess, integrate, track, communicate); UK Modern Slavery Act 2015; French duty of vigilance 2017; EU CSDDD; UNGA Res. 76/300 (2022) · Authors/frameworks: Donnelly (dignity, relative universality); Rorty (sentimental stories, Bosnia 1992); Moyn (Croesus; sufficiency vs equality); Sikkink ("compared to what?"); Van Engeland (universality debate, Rahman apostasy case); Ruggie (embedded liberalism; principled pragmatism); Arendt; Hobbes; Weber; Lemkin; Jonas; Hart's norms table; Reisman (incorporation a national prerogative) · Cases/episodes: Nike sweatshops; Shell Ogoniland 1995 (Ken Saro-Wiwa); cold-open artefacts (UNHCR Amina story, H&M Modern Slavery Statement, Oxfam "What She Makes")
What we did together Justification Tribunal (NordThread/Kesh vignette, five chambers); structured controversy with forced side-switch on HRDD floor/ceiling motion; timeline lab 1919→2024
Three questions to test yourself
  1. Name the three components of claimability and one real situation where each fails.
  2. How did the UNGPs achieve unanimous endorsement where the UN Norms failed?
  3. Explain Moyn's Croesus thought experiment and what it shows about human rights.
→ Day 2 traces how one international norm hardens into national law — choose your yardstick first.
Day 02 — Formation of International Norms & National Legal Frameworks
The ideas at the centre of the day
· Every verdict on rights depends on the chosen comparison (Sikkink)· Tripartism diffuses norms via information, ownership, translation· Progressive realisation and "available resources" build friction by design· Respect is cost-free; protect and fulfil are cost-demanding· Environmental right ran two routes: greening cases, constitutional entrenchment· Proceduralism's limit: full compliance, outcome still missed (2.6°C)· Institutional design trades legitimacy without sanctions against sanctions without legitimacy· Norm hardening is observable in real time (OSH 2019→2022)
The instruments and cases we worked with
Instruments: ICESCR Art. 2(1); Vienna 1993; Aarhus Convention 1998 (information, participation, access to justice); Escazú Agreement 2018; Paris Agreement 2015 (NDCs, ratchet, Art. 2.1(c)); UNGA Res. 76/300 (2022); ILO Declaration on Social Justice 2008; ILO centenary report 2019 (Universal Labour Guarantee); OSH as fifth Fundamental Principle 2022; UN treaty bodies and General Comments; WTO/Marrakesh 1995; Singapore Ministerial 1996 (labour-standards boundary); the "social clause" debate · Authors: Sikkink (Evidence for Hope, ch. 2); Maul (ILO century); Nowak and Bílková (Elgar); Boyle (greening, trilemma); Boyd (92+12 constitutional rights); Addo 1999 ("before" photograph); Tyc (trade and labour) · Cases: KlimaSeniorinnen v Switzerland (ECtHR Grand Chamber, 9 April 2024, Art. 8 and 6(1)); companion cases Carême and Duarte Agostinho dismissed
What we did together Timeline lab void→norm→national law (18-card deck, 1919→2024) with 2035 hardening predictions; Boyle's trilemma mapping; ILO vs WTO institutional relay debate; collective timeline tool (aarhus2026timeline.pages.dev)
Three questions to test yourself
  1. Why do economic and social rights diffuse later and less evenly than civil-political rights?
  2. Describe the two routes to an environmental right and what 2022 showed about them.
  3. What did KlimaSeniorinnen hold, and why is it "greening" rather than a new right?
→ Day 3 asks who enforces international norms inside corporate governance — ESG and reporting frameworks.
Day 03 — Development of Corporate Responsibility Frameworks
The ideas at the centre of the day
· ESG ratings diverge (~0.30 correlation) — no settled measurement object· Divergence lives in measurement (53%) and scope (44%), not weights· CSR, sustainability, ESG follow distinct logics; terminology is unstable· ESG reversed the gaze: what issues do to investment risk· Two theories of change: financial model vs entity model· Financial model won by infrastructure, not adequacy· Outcomes change when systems change, not when initiatives accumulate· Measure outcomes, not programs (the surgeon test)
The instruments and cases we worked with
Frameworks/readings: MacNeil & Esser (financial vs entity model; stewardship paradox); Liang & Renneboog (rating divergence; ~90% non-negative ESG-performance studies); Eccles & Serafeim (performance frontier; four moves; four barriers); Dobbin & Kalev (Getting to Diversity: 3.5%→3.1% finding; what-works evidence table); Nielsen & Villadsen (Danish ESG discourse 2010–2021, 24 large-caps); Tyc ch. 6 (private labour instruments supplement law, not substitute); UNGC/GRI Analysis, SDG target-level mapping; stewardship codes; GRI social standards · Cases/examples: constructed two-verdict ESG rating illustration; Maersk ("licence to operate"), Novozymes, Pandora articulations; Amina in two frames (rights-holder vs financial)
What we did together NordTex A/S workshop: four instruments, eight decisions (materiality, evidence, governance, reporting); plus the What Works Audit (ten initiatives, mechanism + metric required)
Three questions to test yourself
  1. Why do ESG ratings disagree, and why is "weights" the smallest cause?
  2. Contrast the financial and entity models across focus, channel, and metrics.
  3. What did Dobbin & Kalev find about mandatory diversity training, and what works instead?
→ Day 4 puts the NordTex issues under human-rights due diligence and corporate liability law.
Day 04 — From Responsibility to Legal Accountability
The ideas at the centre of the day
· Responsibility asks before harm; accountability asks after harm.· Three gates to remedy: defendant (WHO), forum (WHERE), legal route (WHAT).· Outcomes: blocked, proceeds, settles, judgment, remedy.· Separate legal personality + limited liability split power from liability.· Forum is substantive: evidence, funding, group litigation, enforceability, safety.· Reading discipline: distinguish the holding from the headline.· When one route closes, litigation strategies migrate (ATS → UK tort).· Public oversight claims can become legal evidence.
The instruments and cases we worked with
Five plain-English terms: defendant, forum, jurisdiction, legal route, remedy. · Alien Tort Statute 1789 (28 U.S.C. §1350) — jurisdictional door, not liability code; Sosa constrains actionable norms; rediscovered via Filártiga (1980). · Doe v Unocal — Myanmar pipeline, forced labour; complicity standard (knowing practical assistance, substantial effect); settled before final judgment. · Kiobel (2013) — WHERE: foreign conduct must "touch and concern" US territory. · Jesner (2018) — WHO: foreign corporations excluded as ATS defendants (US corps not categorically). · Nestlé USA v Doe, 593 U.S. 628 (2021) — WHAT conduct: general US corporate activity insufficient. · Vedanta Resources v Lungowe [2019] UKSC 20 — 1,826 Zambian claimants; KCM subsidiary; jurisdiction stage only; arguable direct parent duty; England retained case because substantial justice doubtful in Zambia; settled 2021. Parent-duty routes: policy, management, representation ([52]–[54]); "Embedding Sustainability" report as Exhibit A ([61]). Ownership alone is not enough. · Lubanga (ICC) — criminal-law boundary: ICC prosecutes natural persons, not corporations (Day 5). · Instrument map: litigation (ATS, parent tort) · reporting (UK Modern Slavery Act 2015 — a UK statute) · due diligence (EU & national HRDD, Day 9) · grievance (OECD National Contact Points, Day 10) · criminal law (ICC). · Readings: Joseph; Dodge; Bradshaw; Ojeda; Ford, Gillan & Thein; Wang; Deva & Bilchitz; Liu & Hoskin (Fukushima environmental/interstate contrast).
What we did together Case-routing lab: apply the three-gate grid (defendant, forum, route, decisive gate, decided/not decided, lesson) to Unocal, Kiobel/Nestlé, Vedanta, Lubanga; re-grade the cold-open answers.
Three questions to test yourself
  1. 1. Name the three gates and the gate each of Kiobel, Jesner and Nestlé closed.
  2. 2. What did Vedanta [2019] UKSC 20 decide — and what did it NOT decide?
  3. 3. Which three kinds of parent conduct can make a duty of care arguable, and why does ownership alone fail?
→ Tomorrow business enters armed conflict, where norms, actors and liability routes change again.
Day 05 — Humanitarian Protection and Corporate Responsibility (Leading Responsibly in Conflict-Affected Areas)
The ideas at the centre of the day
· Armed conflict is a legal threshold, not a mood.· In conflict, ordinary business assets become conflict assets.· Recurring question: what does the operation move, enable, finance, legitimize?· Companies need both lenses: IHRL and IHL.· hHRDD analyses impacts on people AND on the conflict.· UNDP: business in conflict areas "will never be neutral".· Decision thresholds: continue, condition, suspend, exit — or repair.· A licence answers one question; end-use diligence answers another.
The instruments and cases we worked with
Legal vocabulary: IAC / NIAC / occupation; civilian, combatant, armed-group member, direct participation in hostilities; civilian object, dual-use asset, military objective (ICRC Customary IHL Rules 7–8); refugee (1951 Convention), IDP, asylum seeker. · Frameworks: UNDP heightened HRDD Guide (2022); UNGPs cause / contribute / directly linked; business-conflict interaction map (context → business function → conflict effect → risk → decision); six-question hHRDD routine (trigger, analyse, map, triage, act, track); Arms Trade Treaty art. 7(3); Geneva Conventions. · Context data: 1 in 6 exposed to conflict (ACLED 2025); 123.2M forcibly displaced (UNHCR 2024); DRC ~75% of cobalt output; resource rents ~13% GDP in fragile economies. · Case capsules: Zyklon B / Tesch & Stabenow (1946–48) — dual-use products and knowledge; van Anraat (NL 2005–09) — accessory to war crimes via chemical precursors, genocide acquittal on intent; Anvil Mining / Kilwa (DRC 2007) — logistics facilitation, acquittals, proof hurdle; Kouwenhoven (NL 2017–18) — Liberian timber, 19 years for arms trafficking and war-crimes complicity; Lafarge Syria (US 2022) — $777.78M guilty plea, material support to ISIS/al-Nusrah; Lundin / Sudan (Sweden) — market repricing before verdict (−7.9% indictment week); Musema (ICTR 2000/01) — superior responsibility of tea-factory director; Facebook / Rohingya — platform accountability gap, unresolved. · Constructive roles: protection/recovery infrastructure; Better Shelter (IKEA Foundation / UNHCR) — good product still needs system governance.
What we did together Country X dossier: board must decide continue / condition / suspend / exit within 48 hours; 200-word executive memo with seven components (recommendation, conflict trigger, business function, conflict effect, legal/ethical concern, immediate controls, evidence request) + 3-min presentation.
Three questions to test yourself
  1. 1. Distinguish IAC, NIAC and occupation, and explain why the threshold changes due-diligence burdens.
  2. 2. Contrast van Anraat and Anvil Mining: what role did knowledge and evidence play in each outcome?
  3. 3. What are the six steps of the hHRDD routine, and what two directions of impact must it assess?
→ Tomorrow asks how companies build the operating system — a due-diligence process that can see, act and prove.
Day 06 — Human Rights Risk Management in Global Business Operations (Due Diligence as Operating System)
The ideas at the centre of the day
· HRDD is an ongoing management process, not an audit event.· Usually an obligation of conduct, not of result.· Operating test: see, classify, act, prove.· Uncertainty should trigger diligence, not excuse it.· Two response tracks: suspend/disengage vs engage and mitigate.· Leaving is also an impact: severity, leverage, dependence, timing.· Leakage: risk moves around the rule instead of disappearing.· A process that cannot be evidenced cannot be defended.
The instruments and cases we worked with
UNGPs Principles 17–21 (HRDD components; cause / contribute / directly linked). · OECD Due Diligence Guidance for Conflict Minerals (2016): five-step engine — management systems, identify & assess, respond, audit chokepoints, report; Annex II model policy; red flags; CAHRA broader than war (instability, repression, institutional weakness); rule — if origin cannot be determined, proceed with due diligence. · OECD Guidelines 2023 — responsible disengagement. · Dodd-Frank §1502 — SEC 3TG disclosure rule (not a sourcing ban); gains (traceability, reduced armed-group control) vs limits (falsified origin, trade diversion, de facto embargo on artisanal miners). · EU Conflict Minerals Regulation 2017/821 — binding importer obligations (3TG, thresholds). · UNSC Res. 1952 (2010) — seeded due-diligence recommendations. · CSDDD — adopted 2024, amended/postponed by Omnibus I; teach direction of travel, not stale timetables. · Readings: Smit et al. 2020 (standard of care / reasonableness); LeBaron 2021 (labour blind spot: wages, purchasing terms); Martínez San Millán 2021 (effectiveness critique); Bartels & Schramade 2024 (human-rights data problem blocks the investor lever).
What we did together Workshop: facts → risk register (risk, source, status, involvement, action, evidence gap) → decision criteria (continue / condition / suspend / disengage) → decision note; debrief tests operational effectiveness and legal defensibility.
Three questions to test yourself
  1. 1. List the OECD five steps and explain why due diligence is a capability, not an inspection.
  2. 2. Which risks demand suspension/disengagement under OECD Annex II, and which call for engagement and mitigation?
  3. 3. What did Dodd-Frank §1502 document as gains, and what is "leakage"?
→ Tomorrow the system meets a live operation: lead a humanitarian organisation under uncertainty and defend the record after landfall.
Day 07 — Humanitarian Business Practices (Simulation, Contracts and Communication)
The ideas at the centre of the day
· Recurring question: delivering help, buying legitimacy, or both?· Exit, voice, loyalty as language for the Day 5 decisions.· Archetypes are objective functions, not personalities.· Every institutional objective makes some harm easier to ignore.· A contract is behavioural design, not just a price.· Two ledgers: money ledger and people ledger.· After every operation, a different story begins.· Compassion can connect — and still be unequal (Fassin).
The instruments and cases we worked with
HumanitSim simulation (aarhus2026gamehumanitarian.pages.dev): mandates, vendor contracts, uncertainty, record after landfall. · Three archetypes: Steward (protect funds; blind spot: delay/unmet need), Humanitarian (minimize social cost = logistics + deprivation; blind spot: exhausting resources), Strategist (fund coverage when marginal benefit justifies cost; blind spot: thresholds make suffering negotiable). · Fan et al. 2024 engine: three objectives × three inventory arrangements (organisation stores / vendor stores / shared) — nine contracts; deprivation cost. · Hirschman 1970 — exit / voice / loyalty heuristic. · Fassin, Humanitarian Reason — crises governed through moral emotions; giver/receiver asymmetry; compassion fatigue. · Atal & Richey 2021 COVID typology: crisis of expertise/logistics, crisis of resources/capital, crisis of the self — whoever defines the crisis defines what counts as help. · Richey & Ponte, Brand Aid — double flow: brands provide aid; aid returns legitimacy to brands. · Five-question campaign grid: SHOWS, NAMES, ASKS, OMITS, RETURNS. · Campaign objects: Ode to Lesvos (Johnnie Walker); Better Shelter clips; Budweiser "Born the Hard Way" 2017 vs "American Icons" 2026 (same emotional machinery, different political risk); COVID sequence — Uber (restraint), Dove (sacrifice as beauty), Airbnb (platform as infrastructure); Danish/European anchors — Novo Nordisk, Maersk Bridge, Armedangels (MSF donation), Zoku (crisis of the self); Ben Affleck Senate testimony — person as brand, "good business". · Ethical communication checklist: voice, resources, politics, delivery, gain.
What we did together Day 5 report-back (exit/voice/loyalty), the HumanitSim simulation with debrief (judgment, contract or luck?), then a communication lab applying the five-question grid to campaign clips.
Three questions to test yourself
  1. 1. Name the three simulation archetypes, what each optimises, and each blind spot.
  2. 2. What are Fan et al.'s three objectives and three inventory arrangements, and why is a contract "behavioural design"?
  3. 3. Apply the five-question grid to a COVID campaign: what crisis is defined, and what does the brand gain?
→ Day 8 asks what leadership culture produces these decisions before the camera arrives.
Day 08 — Building Ethical Leadership to Promote Human Rights (From Personal Influence to Institutional Responsibility)
The ideas at the centre of the day
· Order matters: ethics → authority → evidence → institution.· Moral person vs moral manager: make ethics governable.· Four-layer grid: claim, practice, institution, consequence.· Weber's authority types: traditional, charismatic, legal-rational.· Ladder: notoriety → influence → authority → leadership → governance.· Good intentions can lose against corporate architecture.· Rate the claim, not whether you like the person.· Even admired leaders need rules, records, counter-powers, consequences.
The instruments and cases we worked with
Weber, Economy and Society — three ideal types of legitimate domination; real organisations mix all three; charisma risk: challenge feels disloyal. · Theriono & Lianto 2024 — ethical-leadership vocabulary (moral person / moral manager, after Brown, Treviño & Harrison). · Choudhury 2023 — corporate law's threat to human rights: risk partition (separate personality, limited liability) and incentive pressure can defeat HRDD. · Maak, Pless & Voegtlin 2016 — shareholder advocate (instrumental) vs business statesman (integrative); statesman more necessary where institutions are weak. · Wettstein 2012 — positive human-rights responsibilities beyond "do no harm". · Strand 2024 — Nordic leadership, CEO visibility and institutions. · Analytical anchor: person ↔ institution × projection ↔ evidence (persona, behavioural marks, institutional story, institutional capacity). · Four archetype cases (revealed after rating): A Jensen Huang / NVIDIA — founder charisma+expertise+performance; independent board but exceptional identification; B Jacinda Ardern — empathy in elected office ("They are us"); Christchurch response via legislation, buy-back, Royal Commission; trade-offs of urgent lawmaking; C Roshni Nadar Malhotra / HCLTech — legacy chair as steward; chair/CEO split; non-independent legacy chair; responsible AI also a commercial offering; D Alex Karp / Palantir — mission-driven defence founder ("Our product is used, on occasion, to kill people"); human-rights policy alongside founder voting power up to 49.999999%; 2026 AGM human-rights proposals failed. · HumanitSim connection: archetypes were mandates, not personalities; Moyn (Day 1): rights should not become a ceiling.
What we did together Anonymous live classroom pulse (aarhus2026ethicalleader.pages.dev): Pulse A (ethical-leadership expectations) and Pulse B (influence vs constraint), then per-case legitimacy base + credibility rating (1–5), re-rated after the name reveal with one-sentence justification.
Three questions to test yourself
  1. 1. Define Weber's three types of legitimate authority and give each a corporate echo.
  2. 2. For one case, run the four layers: claim, practice, institution, consequence — what counter-power exists?
  3. 3. Contrast the shareholder advocate and business statesman, and explain when the integrative style becomes necessary.
→ Day 9: when individual leadership becomes governance — a concern must become information, authority, decision and record.
Day 09 — Building Corporate Governance for Human Rights
The ideas at the centre of the day
· Governance route: concern → information → authority → decision → record → follow-up· A visible concern stays ungoverned without authority, obligation, accountability· International law is state-centred; firms hold private political authority (Kobrin)· Voluntary codes harden via practice, contracts, misrepresentation, due-diligence evidence· Directors' duties: shareholder primacy, pluralist models, enlightened shareholder value· Duty usually owed to the company; affected people rarely sue directors· Caremark oversight and UNGP 17 share one architecture: know–build–monitor–respond· Economic complicity: host state as regulator, partner, licensor — and perpetrator
The instruments and cases we worked with
Talisman Energy in Sudan: campaign, ATS suit, exit, possible relocation of harm (Kobrin 2009) · McBarnet 2007: CSR beyond law, through law, for law · Muchlinski 2007: soft commitments become the standard the board must meet · McCorquodale & Neely 2022: directors' duties as HRDD infrastructure; Caremark/UNGP 17 mapping · Siemens/Neubürger: liability for missing group-wide compliance system · Antuzis: exploitation harmed company, contributed to personal director liability · French Duty of Vigilance; EU CSDDD (in force 2024, Omnibus I; transposition 26-07-2028, application 26-07-2029) · Olsen & Bernal-Bermúdez 2023 CHRD: 1,227 claims, >30% state abuse in corporate context · Board screening variables: political past, mineral-rent dependence, union repression
What we did together Moment 4 stress test: build a Vestland Resources governance architecture, then decide stay, pause, re-consult or exit.
Three questions to test yourself
  1. 1. Map the four Caremark-style board acts onto UNGP 17 due-diligence steps.
  2. 2. Why may a board be liable for a missing system rather than the harmful act? Use Siemens/Neubürger.
  3. 3. Talisman exited Sudan under pressure: why might that count as relocating, not solving, the risk?
→ Day 9 built internal reporting to the board; Day 10 asks who outside can see, question and influence that account.
Day 10 — Stakeholder Engagement & Strategic Communications
The ideas at the centre of the day
· Day test: who is heard, who can challenge, what changes?· Stakeholder: anyone whose interests the organisation's activity affects or could affect· Categories: people who cannot opt out, workers/unions, public actors, proxies/experts· Three involvement levels: consultation, participation, steering — match level to risk· Engagement triggers: after harm, before approval, during crisis, preventively· Dialogue cycle: open, sustain, close, process, report back· Engagement must change scope, disclosure, verification or remedy — else information-gathering· Smart mix: national/international × voluntary/mandatory levers combined
The instruments and cases we worked with
Google AI Principles cold open: consulted/participating/steering; affected non-users absent · GRI standards: GRI 1, 2, 3; disclosure 2-29 (who/why/how engaged) · Neighbouring disclosures: 3-1, 3-3, 2-12, 2-16; also 2-25, 2-26 · Bamiatzi et al. 2024: Fortune Global 500 COVID patterns — donors, minimalists, educators, collaborators · DIHR smart-mix framework (Holly & Lysgaard 2021); UNGPs · "Voluntary with a deadline": Norway transparency/right to ask, Finland programme, Denmark/Sweden via EU · Promise-reading test: named people, checkable claims, independent eyes, a consequence
What we did together Nordhavn Roundtable (aarhus2026roundtable.pages.dev): six roles negotiate a Safe City facial-recognition pilot — four clauses (scope, disclosure, verification, remedy), each naming actor, verifier, consequence.
Three questions to test yourself
  1. 1. Distinguish consultation, participation and steering; give a situation justifying each level.
  2. 2. What does GRI 2-29 require an organisation to disclose, and what can GRI not do?
  3. 3. In Nordhavn, why must every clause name an actor, verifier and consequence?
→ Complex technology makes affected people invisible and evidence technical — Day 11 asks what law should require before AI is deployed where risk is serious.
Day 11 — AI in Humanitarian Action and Armed Conflict
The ideas at the centre of the day
· AI is both business output and business capacity with humanitarian consequences· Two domains, one question: what may the machine shape, who stays accountable?· Analytical chain: use case → data → model → decision → human role → consequence → record· Do no harm is a decision rule: anticipate, reduce, pause or refuse· Three deployment gates: data quality, bias/exclusion, consent under power asymmetry· Adoption is organisational capability: understand, test, supervise, stop the system· Human oversight must be real: understand, question, intervene, deactivate, record· Dual-use platforms carry two compliance grammars (civilian vs military)
The instruments and cases we worked with
ICRC 2019 two-domain distinction; ICRC AWS position paper (prohibit unpredictable/anti-personnel AWS; restrict the rest) · IHL judgement tests: distinction, proportionality, precautions, hors de combat (Yordanova 2025) · EU AI Act (Reg. 2024/1689): high-risk duties; Article 2(3) military exclusion; Article 14 oversight/automation bias · NIST AI RMF 1.0: Govern, Map, Measure, Manage · Beduschi 2022 (humanitarian AI risks); Pantiris et al. 2025 (adoption capability); Hirblinger 2026 Digital Peacebuilding · Evidence systems: satellite ML damage mapping (PNAS), VIEWS, ConflictForecast, ACLED CAST · Press file: WCK convoy strike, Kherson drones (Guardian, HRW), surrendering soldiers video, Nasser Hospital camera · UNIDIR 2024/2025: fragmented military-AI governance, make processes complementary
What we did together Move 4 deployment table: groups design an AI deployment agreement that survives crisis.
Three questions to test yourself
  1. 1. Contrast the main failure modes of humanitarian AI versus military AI.
  2. 2. What does the ICRC propose prohibiting versus restricting in autonomous weapon systems?
  3. 3. Why does the EU AI Act's military exclusion not settle a dual-use vendor's obligations?
→ Day 11 ends at capability; Day 12 asks how institutions redesign responsibility when innovation changes the structure of action.
Day 12 — Transformative Innovation, Rights and Collective Action (Closure)
The ideas at the centre of the day
· The canon holds: UNGPs, HRDD, IHL, disclosure, governance, remedy still apply· Falling cost turns tools into infrastructure: scale, dependency, direction, governance· Norms travel, machinery lags, people are missing (implementation gap)· Gap vocabulary: normative, institutional, enforcement, expertise, coordination· Three currents together: democratise capability, concentrate bottlenecks, degrade trust· Fluency becomes authority: confidence mistaken for reliability, judgment delegated early· Complementary rationalities: rights, risk, market, science, democratic, generational· Polycentric governance: no single actor sees enough; coordination is built
The instruments and cases we worked with
Two triads: UNGPs 2011 Protect-Respect-Remedy vs Jones 2023 Principles-Processes-Remedies · Jones 2023: rights as AI baseline; "empty chair" of rights expertise · OHCHR 2024 Mapping Report A/HRC/56/45: gap is implementation, not rulebook · Roberts et al. 2024 risk taxonomy; WHO 2024 large multimodal models, value-chain allocation · Cost-curve evidence: Our World in Data / Farmer & Lafond 2016 · Schot & Steinmueller 2018 three frames; Mazzucato 2018 missions; Ostrom 2010 polycentricity · Foreign Affairs 2026 "AI Trilemma": national vs economic vs societal security · Operating-model stack: UNGP/HRDD + EU AI Act + NIST AI RMF + WHO logic · Hogan & Lasek-Markey 2024: ethics washing and statutory relocation · UN Declaration on Future Generations 2024: reversibility, dependency, residual risk, capacity
What we did together No new in-class activity: bring to Day 13 one horizon signal, your project, and one trusted course tool (six-question project compass).
Three questions to test yourself
  1. 1. Name the five governance gaps and give an AI example of each.
  2. 2. How can one AI system democratise capability, concentrate bottlenecks and degrade trust simultaneously?
  3. 3. Why can binding law still under-deliver rights? Use ethics washing and statutory relocation.
→ Day 13 is integration, not new content: students pitch projects using the course toolkit and their horizon signal.
Day 13 — Capstone
Bring what Day 12 asked: one horizon signal, your project, one trusted course tool. The six-question project compass closes the loop.

Part 3 · The ideas that kept returning

Some ideas appeared on one day and then kept coming back for the rest of the course. This index lists them, with the days where each one lives, so that when you meet one of them in your revision you can quickly find its home.

UNGP pillars & the cause/contribute/directly-linked triageDays 1 · 5 · 6
Three-gate forum routing (defendant · forum · route)Day 4
Obligation of conduct vs outcome; 'reasonable' filled by practice (Smit et al.)Days 5 · 6 · 9
Structure / process / outcome; detection vs incidence; the 0.3 problemDay 3
Norm formation: void → norm → national law; Boyle's trilemma; social clauseDays 1 · 2
The smart mix; hardening on schedule (Holly & Lysgaard)Days 2 · 9 · 10
Consultation / participation / steering; GRI 2-29; represented ≠ presentDay 10
Verification: the four questions; clause test (actor · verifier · consequence)Days 10 · 11
Legitimacy bases; Maak's chain; Wettstein vs ChoudhuryDay 8
Archetypes Steward · Humanitarian · Strategist; exit, voice, loyalty; Brand AidDays 5 · 7
Do-no-harm gates; meaningful human control; EU AI Act; NIST RMF; the carve-out trilogyDay 11
Canon-holds thesis; the five gaps; washing across generationsDay 12
Course tools: Role Compass · timeline · HumanitSim · pulse · Nordhavn roundtableDays 0 · 2 · 7 · 8 · 10

Companion to the portfolio (Assignments 1–3 merged). It replaces nothing: the decks, your notes and the readings remain the substance. Reading relations follow the course's own literature reviews (Blocks I–V).