Each issue states the condition under which its argument fails, and the date that condition resolves. Outcomes are recorded here, whichever way they go.
If the Global Fund for AI Capacity-Building secures endorsement and first pledges by the end of 2027, or the Global Dialogue's second session in May 2027 convenes with published core funding, then this was a start-up lag in a new institution and we read a first cycle too harshly. That is the outcome we want and we will report it with figures attached. If neither has happened — if the May 2027 Dialogue runs again on foundation grants while the World Artificial Intelligence Cooperation Organization's membership grows — then the question stops being about financing and becomes whether a universal forum without a budget can compete with a narrower one that has a product. On present evidence I do not think it can.
The falsifiable claim here is that the summer's incidents reflect a structural gap rather than a bad quarter. If, through the end of 2027, no further frontier-agent incident involves multi-agent coordination — no further covert channel, no further shared protocol between systems meant to be isolated — then July was an artefact of one badly configured evaluation at one company, the hundredfold production-harness figure was the relevant number all along, and this issue over-read it. We will say so. If instead coordination recurs while the research-and-development exemption stands, the AI Office remains understaffed and no jurisdiction requires pre-deployment evaluation, then the record will show that the first documented case of large-scale agent coordination was investigated by three people working unpaid for six days, with no access to the primary model, and that nothing followed. And the specific prediction, for the record: no jurisdiction will require pre-deployment evaluation of frontier models before the end of 2027. The Forecasting Research Institute's panel puts it at 5% by 2030. We expect to be right and would prefer not to be.
If the Nordic AI Act Implementation Network publishes a common Nordic reading of deployer obligations and Norway either introduces the KI-loven or publishes interim supervisory expectations before the end of 2027, the gap described here was a transitional lag competently closed. If neither happens, the craft claim in Issue 02 was a brand claim.
If any frontier laboratory publishes a monitorability metric with a stated halt threshold and an identified decision-maker before the end of 2027, the field closed this gap without external requirements and the argument here was too strong. If none has, the frontier safety case will have been a private assurance for a third consecutive model generation.
If an operator contract is awarded and at least three hubs are running named cohorts by the end of 2027, the design worked and the funding concern here was a misreading of a first tranche. If no operator has been contracted and no hub has opened by then, NAAI joins the National AI Fund as a second AI vehicle without an operational footprint — and the constraint was never money.
If Pakistan’s technology exports keep growing at double digits through FY2026-27 and either a named foreign cloud or compute commitment lands or a published division of labour between the AI institutions appears, the thirteen-month build was a compounding foundation. If exports flatten and no coordination document is published by the end of 2027, this issue over-read velocity as direction.
If blended-finance mobilisation triples to approximately USD 45 billion a year by 2028 and shifts materially toward AI and development transactions with published inclusion criteria, the thesis strengthens; if it does not, capital work moves to the centre of the programme.
If Statnett or the Ministry of Energy publishes an allocation methodology with public-benefit criteria before the end of 2027, the governance gap named here is closed.
If the national AI fund disburses to a published pipeline and training numbers are independently verified before the end of 2028, the drift reading in this issue was wrong.
If the next adoption dataset shows the North–South gap narrowing for two consecutive years, diffusion is self-correcting and literacy policy is not the binding constraint.
If Norges Bank Investment Management publishes the screening methodology, including model scope and human-review thresholds, before the end of 2027, the transparency gap this issue rests on closes.
If a follow-up federal review before the end of 2028 finds AI-specific clauses in a majority of sampled contracts, the diagnosis was a transition artefact rather than a structural gap.
If, by the end of 2028, no national procurement authority in the EU or EEA has published AI-specific contract clauses covering literacy, evaluation and exit, then the tender is not where these decisions are being made.
If no AI cooperation instrument between a Nordic state and a large Global South state reaches signature by the end of 2028, the climate analogue does not transfer.
If a state demonstrably holds all three of control, capability and openness through 2028 without trading one away, the trilemma is not a trilemma.
If Norwegian grid capacity reserved for data centres converts into domestic model training and serving capacity by 2029, rather than into export-oriented colocation, the mismatch closes on its own and no industrial policy was needed.
If a European public buyer adopts inclusion conditions in an AI tender and the resulting deployment shows no measurable difference in who the system works for, then procurement is not the lever this argues it is.
If the Nordic states publish a joint AI governance position by the end of 2027 that survives contact with a divergent national procurement decision, the "brand over craft" reading is too pessimistic and this argument weakens.
If, by the end of 2027, no European supervisory authority has taken an enforcement action or issued formal guidance that turns on Article 4, then literacy was a reporting line rather than a policy instrument and this argument fails.