Increasingly, the world’s oceans are telling us our climate system may be changing faster and more dramatically than expected.

These new insights are made using a vast global network of instruments – from drifting floats and moored buoys to research vessels and underwater gliders – that quietly and continuously feed data to scientists.

Known as the Global Ocean Observing System (GOOS), it provides the fine-grained data that scientists need to detect changes, test climate models and refine projections of future risk.

But now there is rising concern this system itself is at risk – just when the world needs it most.

The hidden system behind modern forecasting

The GOOS is often described as a form of climate monitoring – but it is much more than that. It can best be understood as a network of complementary observing systems, each designed to capture different parts of the ocean in different ways.

Some 4000 autonomous Argo robotic floats sink every ten days down to 2000m depth, before rising to the surface to transmit temperature and salinity profiles to ground stations via satellite.

Underwater gliders target eddies, coastal currents and continental margins where floats cannot go. Elephant seals fitted with sensors collect data beneath polar sea ice in regions no other instrument can easily reach.

Each of these platforms answer questions the others cannot. And ocean observations collected by them now underpin many of the forecasting systems that modern societies rely on every day.

That includes the numerical weather models used to generate daily forecasts, which continuously ingest ocean data to predict evolving weather conditions, as well as newer artificial intelligence-based forecasting systems.

The same is true for hurricane and cyclone forecasts, as well as seasonal forecasting used to anticipate drought, harvests and energy demand. Marine heatwave warnings, sea-level projections and efforts to understand major current systems also rely on sustained long-term observations beneath the ocean surface.

These observations are key for monitoring El Niño climate patterns – including a major event already underway and likely to peak late this year – and major current systems such as the Atlantic Meridional Overturning Circulation.

While satellites can measure surface conditions, they still cannot directly observe the deeper waters where heat accumulates, currents reorganise and the precursors of future weather are already forming.

In short, the GOOS underpins everything from tomorrow’s storm warnings to next century’s climate adaptation plans.

Yet our newly published analysis suggests the system delivering those observations is far more fragile than most people realise.

We found that if observations from a single major contributor, the United States, were withdrawn from GOOS, errors in estimates of how fast the ocean is warming would jump by 163% – worse than randomly losing 80% of all global ocean data.

The reason is largely geographical: US instruments cover every ocean basin and fill critical gaps no other nation currently monitors.

And this is no theoretical concern. Proposed cuts to the National Oceanic and Atmospheric Administration (NOAA) and the National Science Foundation in the United States now threaten exactly this contribution.

Elsewhere, observing systems are also under growing strain, with European programmes facing mounting funding pressure.

In China, scientists and policymakers are trying to build a more resilient national observing effort – but without the resources currently required to fully support it.

A resource the world can’t afford to lose

The total annual cost of operating the GOOS – across all platforms and personnel worldwide – is on the order of US$1.1 billion (about NZ$1.8 billion).

If that sounds expensive, consider that a single major hurricane season can cost the United States hundreds of billions of dollars, while marine heatwaves have already collapsed fisheries and triggered mass coral bleaching around the world.

Compared with the economic damage linked to ocean-driven extreme weather and climate disruption, ocean observation is one of the highest-return public investments available.

The international scientific conference OceanObs'29, to be held in China in three years’ time, will be an opportunity to negotiate a more balanced global observing system – one better aligned with today’s economic realities and maritime interests.

It should also encourage greater scientific cooperation among countries, helping ensure complementary observing networks collectively cover as much of the global ocean as possible.

Maintaining that coverage requires constant renewal.

Argo floats typically last four to five years before their batteries fail. This means they must continually be deployed to prevent gaps emerging across the oceans.

New Zealand plays a surprisingly important role here. Since 2004, the research vessel Kaharoa has helped deploy more than 1,100 Argo floats for international partners across the Pacific and Southern Ocean.

This demonstrates that even smaller countries can use their institutions, expertise and maritime interest to make important contributions.

At the same time, if any one component of the GOOS is removed because of political decisions made in the US or elsewhere, the whole system’s ability to deliver reliable information would degrade.

That would require a rebuild of the system which would prove much more difficult and expensive than the cost of sustaining it today.

More importantly, it could leave the world flying blind into the most consequential transformation of the planet’s climate in human history.

---

The author acknowledges the contributions of Sabrina Speich, John P. Abraham and Lijing Cheng to this article.

---

Kevin Trenberth does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

After weeks of blockades by Iran and the United States in the Strait of Hormuz, it’s clear the narrow waterway is now pivotal to the outcome of the conflict.

The US has begun to escort ships through the narrow passage, but behind the military manoeuvring lies a deeper development: energy security in the Persian Gulf is in a state of profound flux.

As well as the desire by both Iran and the US to control the global flow of oil, gas, helium and fertilisers from the region, the United Arab Emirates (a key US ally) has withdrawn from OPEC in what’s been called a major blow to the oil cartel.

On top of this, Iran has announced plans to introduce tariffs in the Strait of Hormuz as a form of reparations for the damage caused by the war.

If imposed, these tariffs are estimated to be worth between US$40 billion and $50 billion a year to Iran, and would potentially allow it to mitigate the impact of US economic sanctions.

Crucially, tariffs would be a way to cultivate stronger relations with China because they would be denominated in Chinese yuan, not US dollars. This has the potential to significantly alter regional and global power balances.

In fact, such payments have reportedly already been made by vessels going to China, India and Japan, with the Iranian parliament working to formalise the process. (Iran has also begun accepting payments in cryptocurrency.)

50 years of dominance

If Iran can continue to charge these tariffs it could tilt regional influence away from the US towards China and Asia by eroding the historical dominance of the petrodollar.

Essentially, the petrodollar system has seen the pricing and trading of oil in US dollars. The term dates from the 1970s when the US asked Saudi Arabia to exclusively price its oil in US dollars in return for military aid.

This spread across OPEC (the Organization of the Petroleum Exporting Countries), becoming the benchmark of the global oil trade, bolstering the US dollar as the global reserve currency and underwriting US power.

Oil-producing nations amassed huge petrodollar surpluses – too much to invest only in their own economies – which were funnelled or “recycled” back into US securities and stocks, and other countries’ sovereign wealth funds.

They have become the primary source of revenue for OPEC members, as well as non-member oil exporters Qatar and Norway. This ties these countries to Washington and gives the US significant financial leverage in global affairs. The flow of petrodollars helps finance US deficits and reduce US borrowing costs.

A new paradigm?

If major regional players such as the UAE, Bahrain, Qatar, Kuwait and Saudi Arabia pay Iranian tariffs in “petroyuans”, economist Antonio Bhardwaj has said, it would mark:

the systematic erosion of the petrodollar system and the emergence of the petroyuan as a credible, institutionally embedded alternative framework for settling global energy transactions.

It’s a sizeable “if”, but the introduction of tariffs would also pose a dilemma for countries that supported Iran in the conflict (implicitly or explicitly) and those that didn’t.

As internatinoal relations analyst Pakizah Parveen has written, we would see the emergence of:

a bifurcated global oil market: barrels from compliant parties would move through Hormuz in yuan. In contrast, non-compliant parties would incur significantly higher costs in dollar-denominated barrels.

Such a choice would affect major US allies such as Pakistan, South Korea, Japan and the Philippines, all of which have faced severe economic pressures as a result of the upheavals in the Gulf and Middle East.

Paying tariffs in petroyuan would draw them towards China and play into Beijing’s narrative of being a reliable and more stable economic force. It also mirrors Russia’s request for payment in yuan for its oil since 2025.

Decline of the petrodollar

It would be premature to argue Iranian tariffs will lead to a general “de-dollarisation” of the world economy. But they may be a step towards a devaluing of the US dollar.

By extension, any move by other countries away from the US dollar is a move away from dependence on the US financially and politically. It would also aid China’s push to internationalise the yuan.

For the first time since 1996, global central banks hold more gold in their reserves than US debt securities. The BRICS group of countries may move further away from US influence, with China, India and Brazil having all reduced their US holdings in 2025.

Overall, Iranian tariffs denominated in yuan would be another sign of an emerging multipolar world in which US preeminence is no longer a given. It would mean more strategic flexibility for all countries, great and small, but also more uncertainty.

Chris Ogden is affiliated with the Foreign Policy Centre in London.

With the lid now lifted on Budget 2026, many small and medium New Zealand businesses will be poring over the detail to see what it has in store for them.

Many may come away disappointed. With the government having been upfront about its spending constraints, this budget was never likely to deliver a large new package for small firms.

Instead, the budget delivers a mix of smaller compliance changes, infrastructure spending and energy transition support. It also includes funding for advisory services, digital systems, early-stage capital initiatives and small-business capability programmes, alongside changes to research and development support.

It might be asked whether this mix of small initiatives adds up to anything that will lift productivity. But aside from offering direct support, budgets can also act as useful signals from government.

On this test, this budget appears strongest on encouraging restraint and resilience in a difficult economic environment, even if it is less clear how ordinary small and medium firms are expected to become more productive.

That is important because New Zealand’s productivity problem will not be solved by large firms, infrastructure projects or high-growth startups alone. It also depends on whether thousands of everyday businesses can lift capability, adopt technology and improve margins.

There is an important distinction here. Business support helps firms navigate rules or cope with pressure. Productivity policy helps firms change how they create value. If we look closely, there are measures in the budget to assist in both areas.

The hidden small-business budget

Several budget initiatives are relevant to small firms.

Simplified fringe benefit tax rules for private motor vehicle use will be welcomed by owner managers, trades businesses and service firms. Removing detailed logbook requirements may not sound transformational, but compliance time is time taken away from customers, staff, sales and cash flow.

The $1.2 billion gas transition loan guarantee scheme may help some energy-intensive firms shift to alternative energy sources. Infrastructure spending, trades training, advisory services and digital tools also have practical value.

But these are mainly support measures. They may help firms cope, navigate or adjust. The harder test is whether they help small firms become more productive.

Some of the most interesting measures are less glamorous.

Customer and product data sharing is one example. The budget documents include support for an economy-wide framework to enable more secure data sharing between businesses, with open banking as an early focus.

For smaller firms, access to finance is shaped by information. A viable business may have loyal customers and solid cash flow but still struggle to demonstrate its risk profile to lenders. Better data sharing might make lending, switching providers and cash flow management easier.

Electronic invoicing, or eInvoicing, is another example. For a small firm, a late invoice can mean drawing down an overdraft, delaying a supplier payment or chasing accounts after hours. While eInvoicing will not solve late payment by itself, combined with stronger payment discipline it can reduce errors, speed processing, and improve cash flow visibility.

Procurement is another overlooked lever, although the budget is less explicit here. The documents include funding for procurement leadership and improving the experience of businesses interacting with government.

For an innovative small firm, a government contract can provide revenue, credibility and a first reference customer. The budget does not present procurement as an innovation strategy, but that is the opportunity.

If New Zealand wants more innovative small firms, government can buy innovation as well as fund it.

Startup policy is not small-business policy

Budget 2026 also includes support across the wider business ecosystem, including startups and early-stage capital.

New Zealand needs ambitious startups and deeper early-stage capital markets.

But startup policy is not the same as small-business policy. Most small and medium businesses are not seeking seed capital or pitching venture investors. They are established firms trying to manage costs, retain staff, respond to weak demand, adopt technology, improve systems and lift margins.

The Research and Development Tax Incentive changes show the same distinction. In-year payments and greater discretion around late filings may improve cash flow and administrative certainty for eligible businesses actively engaged in research and development. But this is not a broad measure for most small businesses.

The package also reduces the cap on eligible expenditure for non-administrative internal software development from $25 million to $3 million a year. The government says this will better target support toward activities that generate wider spillover benefits.

But some internal software development is central to innovation and productivity. At a time when firms need more sophisticated digital tools, narrowing support for this kind of software development sends a mixed signal.

Building blocks, but no blueprint

There are things in the Budget for small firms. The issue is whether the measures add up to a credible blueprint for lifting productivity.

A small business under pressure does not experience government policy as a set of discrete measures. It experiences the cumulative effect. Does it become easier to access finance, adopt technology, reach customers, draw on expertise and meet compliance requirements without sacrificing productive time?

Business understands fiscal constraints, but still looks to government for signals of vision and coherence.

That is the missed opportunity. In a constrained fiscal environment, the government did not need a grand small-business package. It needed a clearer growth story for the firms that make up most of New Zealand’s business economy.

Rod McNaughton does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Summer heatwaves are currently receiving a lot of attention in Europe because they now cause more deaths than floods or storms.

But winters are also warming. While they are generally less deadly, they influence and disrupt human and natural systems in many subtle ways.

Aotearoa New Zealand has experienced a particularly warm start to this winter, with record high June temperatures in the capital and warm conditions across the country.

Many will welcome the unseasonably warm weather, but milder winters have a range of impacts, especially for plants and insects.

Extra winter growth, but loss of carbon

In forests, warmer temperatures can extend the growing season of trees.

Usually, many trees are dormant during winter as conditions are too cold for growth. But our ongoing measurements of kauri tree growth in Auckland indicate trees have continued to grow throughout recent winters.

One might assume a longer growing season would increase carbon uptake and storage in trees. However, overall carbon changes are actually negative because warmer temperatures also increase respiration, which returns more carbon to the atmosphere.

In Aotearoa, few plant species lose their leaves in winter. But according to traditional Māori knowledge (mātauranga Māori), flowering time has changed and fruit biomass has declined with warming in forests of the central North Island since the 1950s.

This has had a negative impact on the numbers, breeding rates and health of kererū (native wood pigeons) and has reduced nutrient cycling in the soil.

Risk of new invasions

Insects are also very sensitive to winter temperatures.

Like trees, many insects have a dormant period during colder months. Some insects from warmer climates have established as pests in Aotearoa, but they usually struggle to survive cold conditions. As winters warm, the numbers of species able to get through the cold season is increasing.

For instance, in temperate climates such as in New Zealand, wasp colonies have a strong seasonal cycle. Wasp numbers increase during spring after the queen emerges from overwintering and lays eggs. The workers expand the colony during summer but when temperatures drop in autumn, most of them die off.

However, in warmer conditions, sightings of winter-active workers have increased in Aotearoa. This means a warming climate will likely lead to higher wasp numbers and increased ecological and economic impacts.

There are a range of other invertebrate pests that may become more problematic in natural systems, plantation forests and agricultural and horticultural settings as winters warm. This includes rising numbers of parasites of sheep and cattle, more insect pests in plantation forests, increased risk of overwintering of the Queensland fruit fly and bigger range sizes for mosquitoes and ant and cattle ticks.

Shrinking alpine refuge

New invasive plant species from subtropical regions may also be able to establish or expand their ranges and shift into the alpine zone.

Similarly, the upward expansion of invasive mammals will reduce the availability of refuge areas for native birds, including the endangered rock wren.

Known as “thermal squeeze”, the movement of rats and stoats to higher elevations reduces the availability of safe spaces for large alpine birds such as the kea, exacerbating the risk of extinction.

The alpine zone is especially vulnerable to winter warming because plants and animals living there are highly adapted to the specific environmental conditions and are often poorly prepared for invasive predators or competitors.

Horticultural winners and losers

In the horticulture industry, cold winter nights are important as a trigger for spring flowering. Economically important fruits such as apples, avocados and kiwifruit may not flower well and have poor-quality fruit under future climates.

Potatoes and onions are also sensitive to warming conditions because heat stress reduces the quality of tubers and produces smaller bulbs, causing lower yields of both crops.

Plant breeding and gene technologies offer opportunities to develop fruits and vegetables that are better prepared for a warmer world.

And there is some good news in other areas, including that flea infestations are predicted to decline in regions where warming is associated with drying. There may also be opportunities for the establishment of new crops, such as bananas.

As the climate continues to warm, there is more to learn about the impacts and options for adaptation in Aotearoa. Research needs to focus on finding solutions for native species and primary industries because healthy ecosystems are essential for a healthy economy and thriving communities.

Cate Macinnis-Ng has received funding from The Royal Society Te Apārangi, MBIE and Te Pūnaha Matatini. She is a lead author for the IPCC Assessment Round 7 Working Group II report.

Stopping judges from considering “good character” references when sentencing sexual offenders – as New Zealand’s National Party has pledged to if re-elected – may sound like a niche legal reform.

But it targets a real and longstanding issue in the country’s criminal justice system – and one that has drawn renewed public attention and debate over recent months.

In the courts, character references are typically used to show a defendant has no prior history of similar offending. Lawyers may point to the absence of previous convictions and present supportive letters describing the assault as out of character.

Such evidence can be introduced by defendants convicted of – or pleading guilty to – sexual offending in a bid to reduce their sentence, alongside other mitigating discounts that judges can apply under the Sentencing Act.

National argues that its proposed reform, which comes alongside a separate petition and campaign, would lead to tougher sentences and stop offenders benefiting from their personal reputation or social standing.

Some defence lawyers, however, have argued judges already treat such evidence cautiously in serious sexual offending cases, and warn that removing it entirely could undermine the principle that courts should consider all relevant circumstances at sentencing.

In any case, the move would represent a meaningful change. But the discussion also raises wider questions about New Zealand’s sentencing framework itself – particularly when it comes to how much discretion judges are presently given.

The problems with ‘good character’ references

As Justice Minister Paul Goldsmith noted when announcing the move last month, good character references are often used to argue the offending was “the exception, not the rule”.

In sexual violence cases, this can involve employers, relatives or community figures portraying the defendant as an otherwise respectable person who made a one-off mistake.

Rape cases, particularly, illustrate the flaws in this type of reasoning. Presenting evidence of a defendant’s good character can reinforce the myth that there is a meaningful distinction between a “real rapist” and someone who has merely committed rape.

This framing also risks minimising the seriousness of sexual violence and obscuring the reality that most rapes are committed by someone known to the victim, often in private places and with little or no physical force.

Another problem is that this evidence can be deeply retraumatising for victims, who may have to watch the sentencing judge consider – and sometimes even credit – claims that the assault was less serious, or rather something more akin to a misunderstanding.

If the policy choice is between continuing to treat prior “good character” as mitigation in sexual violence cases, or scrapping it, the latter would arguably appear the sensible call.

But abolishing this single mitigating factor from the Sentencing Act – at least as it applies to sexual offences – still leaves many other issues within the legislation to address.

The case for wider reform

In another development last month, an advisory group was established to bring lived experience and leadership expertise into government decision-making around family and sexual violence prevention.

While this marks an important step, overseas experience suggests New Zealand could go much further in reforming its sentencing system.

Countries including the United States, the United Kingdom, Canada and many Australian states, for instance, use sentencing commissions to develop formal sentencing guidelines.

These bodies draw on expertise from criminology, psychology, statistics and criminal law to analyse research and sentencing data, then produce guidance on how different offences and offenders should be sentenced.

The resulting guidelines help to eliminate disparities across offences, offenders, judges, and geographic regions, while also ensuring transparency in sentencing policy. They also tend to rely more on evidence and risk-based assessment than on broad and often ambiguous factors gradually developed through court decisions.

By contrast, two-decade-old Sentencing Act appears antiquated.

Aggravating and mitigating factors referenced within the legislation are often intuitive, vague and morally framed, rather than being clearly defined or grounded in evidence.

Importantly, they also provide little meaningful guidance for how judges should apply them consistently across cases involving different levels of harm, premeditation or remorse on the part of the offender.

Leaving sentencing judges with such a high level of unguided discretion risks allowing implicit biases – which all people possess – to influence sentencing decisions.

The result is that subjective assessments about who seems dangerous, remorseful or respectable can end up driving sentencing decisions, rather than being based on consistent, evidence-based assessments of harm, proportionality and risk to public safety.

Removing “good character” mitigation in sexual violence cases may therefore be worthwhile. But if New Zealand wants a better sentencing system, much broader reform is required.

Carrie Leonetti does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

From personal health portals to AI assistants that draft emails, the digital age has simplified endless everyday tasks.

But for many older New Zealanders, the rapid march of technology has helped build a wall rather than open doors. Navigating online forms, changing apps, disappearing face-to-face services and the constant threat of scams can be daunting.

There is a term for this unease: technostress. Once used to describe the anxiety and frustration felt by workers, it has more recently been applied to older populations struggling in our digital-by-default world.

While older people’s overall digital engagement has grown over the years, about half of over 50s feel they are being left behind by modern technology.

Amid a planned public sector shake-up that would further digitise services, more than 40% of people older than 60 face barriers for accessing online government information.

More than ever, digital inclusion has become a necessity for older people to access essential services and maintain social connections. Without it, there are serious implications for their psychological, social, cognitive, physical and financial wellbeing.

Our newly published research, based on interviews with 23 people aged over 65, reveals a complex relationship with technology: one that can support independence, but also create new sources of stress and exclusion.

A double-edged sword

The experiences of those we interviewed varied widely. Some used technology very little – perhaps just for texts or phone calls – while others relied upon it heavily for daily chores and work. One study participant spoke enthusiastically about using an AI assistant to support her creative writing.

But regardless of how tech-savvy they were, all felt that keeping up with digital change was a never-ending but necessary challenge. This was especially apparent for those who used tech during their working years but found fewer resources available to upskill in retirement.

Another common theme was feeling targeted by scammers due to their age. For people living on a single income or pension, the financial risk of falling victim to a scam could be devastating and put them off going online.

Broadly, we found technology to be a double-edged sword for older people. For those who felt digitally included, it helped strengthen relationships through sharing photos and videos with family overseas and provided useful access to health information.

For those who felt shut out, technology became a source of distress, frustration and feelings of incompetence. They described struggling with online pension applications or having to relearn familiar software after unexpected updates completely changed the interface.

Some felt the accessibility features built into everyday digital devices were inadequate for their physical needs, causing them to abandon tasks because of eye strain or frustration.

Others felt digital technologies were not culturally responsive, reflecting a predominantly Western worldview. Common errors, such as the mispronunciation of te reo Māori names, could deepen feelings of exclusion and cultural invalidation.

Ageism and equity

Nearly all participants felt digital technologies were not designed with older people in mind, believing they were not viewed as a priority market by mainstream technology companies.

Yet many still blamed themselves for struggling to keep up. Some also described dismissive or impatient responses when seeking help, reinforcing feelings of frustration and inadequacy.

This may suggest a problem of digital ageism: the assumption that older people use technology less because they are unable or unwilling to engage with it.

In reality, meaningful digital participation depends on much more than willingness. It requires people to have the motivation, skills, confidence, access, trust and support needed to engage safely and effectively.

In this context, the challenge is not about age but equity. Fortunately, many organisations and individuals across Aotearoa New Zealand are working diligently on this issue, from advocacy groups to library-based digital skills programmes.

Some local companies have introduced equity products, such as age-friendly digital tablets and lower-cost mobile plans for pensioners. These efforts are essential, but more is needed.

More recent interviews with people working in the field suggest a need to dispel the myth that digital exclusion will disappear as older people increasingly become “digital natives”.

Instead, digital inclusion should be viewed as a fundamental right rather than a luxury in the hands of for-profit companies. This will require targeted policy, increased collaboration across sectors, and active involvement of older people as equal partners in digital design, testing and decision-making.

New Zealanders cannot benefit from even the most well-intended developments, such as telehealth services, if even one aspect of digital inclusion is lacking.

Without deliberate action, the shift to digital services risks widening the very inequities it is intended to reduce.

The author acknowledges the contributions of Rosie Dobson, Cinnamon Lindsay Latimer, Judith McCool, Robyn Whittaker and Vili Nosa to this research.

Melanie Stowell receives a doctoral scholarship from the University of Auckland and support from the Centre for Co-created Ageing Research.

Three months into the war in Iran, the largest disruption to New Zealand’s oil supply in living memory appears to have done nothing to change the government’s approach to transport.

Budget 2026 spends more on roads, less on rail and nothing on walking or cycling. It also raids bus decarbonisation to fund a review of pipes.

Finance Minister Nicola Willis warned ahead of budget day there would be no extravagance this year – and the government’s now-released spending on transport and infrastructure confirms it.

The fiscal logic underpinning its priorities assumes the Middle East conflict and its impacts on fuel markets will ease in time for a projected operating surplus by 2028/29.

Whether that happens remains to be seen. What is apparent, however, is that this budget does little to reduce the reliance of New Zealand’s transport system on fossil fuels – and instead reinforces it.

Roads of ‘national sacrifice’?

Te Waihanga, the government’s own infrastructure commission, has warned for years that the Roads of National Significance programme is unaffordable.

Budget 2026 commits NZ$1.773 billion for a single one of them, Cambridge to Piarere, to be supplemented further from the National Land Transport Fund. Another $400 million has been set aside for State Highway resilience.

Combined, that’s $2.173 billion, about three quarters of all newly allocated transport capital spending, channelled to roads and highways.

Metropolitan rail gets $106.9 million for “one year of additional capital” to clear overdue renewals in Auckland and Wellington. A further $598.2 million in capital funding and $477.1 million in operating funding is tagged for the national rail freight network.

Among the spending across all infrastructure, not a single cent is specifically earmarked for walking and cycling.

While public transport users will get no fare relief and no expanded service, the Treasury floats the option of deferring the planned 12-cents-per-litre increase to Fuel Excise Duty and Road User Charges scheduled for January 2027.

Each six-month deferral, the Treasury reports, would cost about $300 million in foregone revenue.

For comparison, $300 million for six months of relief at the pump is roughly the entire annual fare revenue generated by every public transport service in the country combined.

The same $300 million spent on free or subsidised fares would deliver a full year of relief to people who do not own cars, cannot afford to run them, or have chosen not to.

The government has also signalled it’s willing to forgo road revenue, while telling the country an equal or lower expenditure to reduce public transport fares is too expensive.

The budget does include a Public Transport Fuel Costs Support initiative, also linked to the Middle East conflict. However, the amount has been withheld on the grounds of “commercial sensitivity and negotiations”, making it difficult to assess the scale of the support being offered.

To help fund its transport package – comprising $456 million in operating spending and $2.708 billion in capital – the government is reclaiming money from existing programmes.

That includes clawing back $12.2 million from initiatives such as the now-discontinued local-road Crown Resilience Programme. It is also redirecting $170 million from the current Rail Network Investment Programme toward new metro and freight renewals.

Most pointedly, it draws $2.5 million from the Public Transport Bus Decarbonisation Fund and the Older Drivers Licence Holder Subsidy, redirecting both to a review of underground infrastructure assets in another portfolio.

While the amount is relatively small, taking money from one of the cheapest near-term ways to reduce urban transport emissions and fuel use is still notable.

Long-term risks, limited investment

Outside transport, the fiscal picture for infrastructure is no less tight.

There is no named freshwater, stormwater, wastewater or drinking-water capital package. The government allocates $29.7 million to oversee development levies, $294 million in operating spending and $13.4 million in capital spending for resource management reforms, and $400 million in tagged contingency for council housing-growth incentives.

Treasury’s own forecasts show core Crown spending on water infrastructure falling from $127 million in 2026 to $5 million per year by 2028. After a decade of weather losses already surpassing $15 billion, the country’s stormwater and flood resilience response is a dwindling budget line.

There is one new acknowledgement of the oil supply problem itself: $150 million is allocated to support additional fuel supply, including fuel reserves. The government has chosen, again, to insure people against an oil-shocked world rather than reduce exposure to it.

The Budget papers also identify several major future fiscal pressures tied to transport infrastructure.

Transport project funding is logged as a $5 billion to $10 billion risk. National Land Transport Fund sustainability is another $2 billion to $5 billion. Most of these costs sit outside the forecast horizon, which is to say the next government will pay them.

Arguably, this could not be called a transformative transport or infrastructure budget. Rather, it is a fiscally cautious package that continues to prioritise road investment over measures aimed at reducing fuel dependence or expanding transport alternatives.

Public transport subsidies and cycleways that could help reduce exposure to oil shocks are largely absent. Buses that might otherwise have been electrified will continue relying on imported diesel at a time the budget itself acknowledges fuel supply vulnerabilities.

Meanwhile, Treasury forecasts show stormwater infrastructure spending falling sharply over the coming years despite the growing costs of extreme weather events.

New Zealand’s long-term transport and infrastructure challenges are well understood. Budget 2026, however, suggests the government has chosen, at least for now, to prioritise fiscal restraint and existing transport priorities over much-needed structural change.

Timothy Welch does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

The world relies on a modest number of countries to keep watch over the ocean. That arrangement is starting to fail. Europe and Asia must now decide whether to let the system unravel, or to take it up together.

Right now, in every ocean basin on Earth, a global network of instruments measures the state of the sea.

Research ships steam along oceanographic transects from surface to seafloor. Anchored buoys watch the tropical oceans for the first signs of El Niño or tropical cyclones and take the pulse of the thermohaline circulation. Some four thousand autonomous floats sink every ten days to two thousand metres before rising to transmit temperature and salinity to ground stations via satellite. Underwater gliders patrol continental margins, and drifting buoys ride the surface in the most remote waters. Hundreds of elephant seals carry miniaturised sensors beneath the polar sea ice…

Together, this network produces invaluable information that allows societies to anticipate and respond to a changing ocean and weather conditions, and protect the ocean in return.

It is also far more fragile than most people, and most governments realise. A new study published in Nature Climate Change has measured for the first time just how fragile the ocean watch network is.

The result is alarming. If observations from a single major contributor, the United States, were withdrawn from the Global Ocean Observing System (GOOS), the errors in our estimate of how fast the ocean is warming would jump by 163 percent. That is worse than randomly losing 80 percent of all global ocean data. The reason is geographical: US instruments cover every ocean basin and plug critical gaps that no other nation currently fills.

This is not a theoretical concern. Proposed cuts to National Oceanic and Atmospheric Administration (NOAA) and the National Science Foundation in the United States now threaten exactly this contribution. And the situation is barely better on the other side of the Atlantic.

The pressures are not confined to one side of the Atlantic, nor to the West. In China, scientists and policymakers are working to build a more resilient national contribution to ocean observation, but without the resources the moment requires. The marine monitoring system the world relies on is under strain almost everywhere.

An observing system, not a programme

Public conversations about ocean observations often focus on Argo floats.

Each Argo is essentially a sealed cylinder of pressurised electronics with a clever buoyancy chamber: it floods with seawater to sink and is evacuated to rise again. These autonomous robots have transformed ocean science this century.

However, Argo is just one component of GOOS and the complementarity of its parts matters.

• Argo profiles the upper two kilometres of the open ocean.

• Research vessels go deeper: GO-SHIP cruises survey from surface to seafloor along long repeated transects, providing the high-precision reference measurements that calibrate every other instrument and help validate climate models.

• Moored buoys deliver continuous time series critical for monitoring El Niño, the Atlantic Meridional Overturning Circulation, and the conditions in which tropical cyclones form.

• Underwater gliders target coastal currents, eddies and continental margins that floats cannot resolve.

• Elephant seals carry sensors into under-ice regions of the polar oceans that no other instrument can reach.

Each platform answers questions the others cannot.

Remove any one of these ocean watch components, and the observing system’s ability to deliver reliable information degrades not in proportion to the volume of data lost, but in proportion to where the gaps appear.

What this network actually delivers

The Global Ocean Observing System is too often described as “climate monitoring,” but it does so much more.

Every operational weather forecast is built on these data. The numerical weather prediction systems run by the European Centre for Medium-Range Weather Forecasts, by Météo France, and by every other major weather service ingest ocean observations many times a day.

Without them, forecasts drift quickly out of skill.

The new artificial-intelligence-based forecast systems Pangu-Weather and GraphCast, despite their impressive performance, rely entirely on the same observational stream.

AI does not replace observations; it depends on them.

Sub-seasonal to seasonal forecasting, to help anticipate harvest seasons, energy demand and water availability weeks to months ahead, depends critically on knowledge about subsurface ocean heat and salinity.

Tropical cyclone track and intensity forecasts, central to early warning and evacuation decisions, depend on ocean heat content beneath the surface, not just sea surface temperature, because hurricanes draw their explosive energy from the warm layers down to at least 200m depth.

Marine heatwave warnings, now used routinely by fisheries managers worldwide, are impossible without sustained subsurface observation.

Sea-level projections used to design coastal infrastructure require decades of consistent measurements, and salinity adds the density information essential for determining all ocean currents, including the AMOC.

In short, GOOS underpins operational services from tomorrow’s storm warnings to next century’s adaptation plan. It is not a luxury but essential.

Why models and artificial intelligence alone cannot save us

There is a persistent misconception, amplified by the rise of AI, that sufficiently advanced models can substitute for direct observations. They cannot.

Every forecast model, whether traditional or AI- based, relies on data assimilation: a continuous nudged adjustment of the simulation toward real-world measurements. An AI model trained on a richly observed past will perform poorly in a sparsely observed present. In a world of rising extremes and shifting ocean states, historical patterns become less reliable.

An observation not made is lost forever. Satellite measurements of the sea surface cannot tell us what is happening hundreds or thousands of metres below, where heat accumulates, currents reorganise, and the precursors of the next season’s weather are already forming. To see beneath the surface, we need instruments in the water.

The cheapest insurance we have

The argument that ocean observation is too expensive collapses on contact with the numbers.

The total annual cost of the global system, across all platforms and personnel, runs on the order of one billion euros worldwide. The European share is a fraction of that.

Extreme weather events linked to ocean conditions caused tens of billions of euros in damage across Europe in 2024 alone.

A single major North Atlantic hurricane season can cost the United States hundreds of billions of dollars. Marine heatwaves have collapsed fisheries worth billions and triggered mass coral bleaching on every reef system on Earth. Failed seasonal forecasts cascade through agriculture, energy and humanitarian response, with consequences rarely tallied.

Every euro spent on ocean observation returns many times its value. It is one of the highest-return public investments available.

Europe’s choice

Europe needs to treat ocean observations as critical infrastructure, equivalent to satellite navigation or meteorological services. That means stable, multi-year funding for the operational backbone of the system: the floats, ships, moorings, gliders and data centres that process and deliver the data.

France has the second-largest Exclusive Economic Zone in the world but contributes around 5 percent of global ocean temperature profile data. Present in the Atlantic, Pacific and Indian Oceans, France has five departments and regions and seven overseas collectivities, which are home to 2.7 million French citizens.

Australia contributes more than three times as much.

The European Union contributes about 12 percent, less than a quarter of the American share. Europe, and France in particular, should substantially increase its contribution.

OceanObs’29, the decadal international conference set to be held this time in China, is an opportunity to negotiate a more balanced global system, reflecting economic capacity and maritime interests rather than historical accident.

Europe-China scientific cooperation should increase, as they have largely complementary observing footprints. Together, they would cover much of the global ocean.

The narrowing window

The danger is a gradual erosion of the information on which a growing share of human activity and the blue economy now depends.

Cyclone warnings become less reliable, seasonal forecasts less skilful, sea-level projections less precise. Each loss maybe individually tolerable. Together, they amount to flying blind into the most consequential transformation of the planet’s climate in human history.

The ocean observing system is a planetary public service, built over decades by many nations. France and Europe possess the institutions, expertise and the maritime interest to play a far larger role.

What is missing is the political decision to act, while the system can still be sustained. The loss of collaboration among nations would force a rebuild far more difficult and expensive than sustained investment in what already works.

The window remains open but is narrower than it was.

---

A weekly e-mail in English featuring expertise from scholars and researchers. It provides an introduction to the diversity of research coming out of the continent and considers some of the key issues facing European countries. Get the newsletter!

---

Sabrina Speich a reçu des financements de l'ERC, EU Horizon 2030, CNES TOSCA et ANR. Sabrina Speich est présidente du comité d'experts du "Ocean Observations for Physics and Climate" des programmes UN GOOS et GCOS.

John Abraham, Kevin Trenberth et Lijing Cheng, ne travaillent pas, ne conseillent pas, ne possèdent pas de parts, ne reçoivent pas de fonds d'une organisation qui pourrait tirer profit de cet article, et n'ont déclaré aucune autre affiliation que leur poste universitaire.

The US state of Florida has filed a lawsuit against OpenAI and Sam Altman, alleging the tech giant and its CEO put profit over public safety with its flagship artificial intelligence (AI) product, ChatGPT.

The lawsuit, filed in Florida state court on Monday local time by Florida’s attorney general James Uthmeier, is one of the most significant enforcement actions brought by a state attorney against an AI company to date.

It comes as OpenAI and other big tech companies are embroiled in a growing number of legal cases related to the alleged harm their products have caused.

Six key elements

The complaint opens with a screenshot of OpenAI’s own parental-control page, which states that ChatGPT was “built with safety in mind”. Then, in a standalone paragraph, the State answers with two words: “Not so”.

This signals the central allegation of the case: that OpenAI sold ChatGPT to the public as safe and reliable, while knowing it could cause serious harm

More specifically, there are six key themes to Florida’s case against OpenAI. The first is that the company engaged in deceptive safety marketing, assuring parents the platform is safe for teenage use, while not clearly disclosing that ChatGPT can be wrong.

Second, despite OpenAI’s marketing, ChatGPT is unreliable. A 2025 study, for example, found AI assistants, such as ChatGPT, misrepresent the news roughly 45% of the time. Similarly, despite marketing suggesting ChatGPT can handle financial affairs, ChatGPT has failed in meeting basic accounting standards and provided incorrect tax advice to users.

The third element of the case is the public safety threat. The danger to young people in particular is illustrated by the tragic story of Adam Raine, a 16-year-old who died by suicide in April 2025 after engaging in long conversations with ChatGPT. When Adam expressed suicidal thoughts, ChatGPT responded that it “won’t try to talk you out of your feelings”. It helped Adam plan a “beautiful suicide” and even offered to write his suicide note for him.

Why would a product behave this way? Because, Uthmeier argues, it was built to.

OpenAI designed ChatGPT to be highly agreeable, to say “yes” roughly ten times as often as “no”, according to a Washington Post review of 47,000 conversations. This forms the fourth element of the case – commercial exploitation through sycophancy. In other words, ChatGPT optimistically parrots back users’ responses in order to to manipulate them into deeper conversations, regardless of truth or safety.

But according to the lawsuit, even ordinary use carries a cost: it weakens people’s brain activity and critical thinking skills (also known as cognitive atrophy). This is the fifth element of the case.

The sixth and final element is knowledge – specifically, the knowledge of Samuel Altman. According to Uthmeier, since at least 2023, OpenAI’s own documents warned that the model could coach people on committing crimes, but Altman overruled the safety staff.

These six elements paint a picture of a product marketed as safe, engineered to be addictive, and known by its own makers to be dangerous – yet sold to us, anyway.

Altman is at the centre of that picture. The complaint reconstructs his career and reaches for an April 2026 New Yorker investigation and testimony from the recent legal battle between Elon Musk and OpenAI to depict a man, who in Uthmeier’s telling, repeatedly chose speed over safety.

That is why Uthmeier is asking the court to hold Altman personally liable for “his utter disregard for the risk to human life”.

Pay for past harms

Uthmeier is asking the court to declare that OpenAI broke the law, then to order the company to stop – permanently – its unlawful practices.

He wants the company barred from collecting children’s data without parental consent and the safeguards that should come with it, and barred from misrepresenting or staying silent about ChatGPT’s risks.

On top of the injunctions, the state is seeking civil penalties of up to US$10,000 per violation for OpenAI’s alleged wilful violation of the the Florida Deceptive and Unfair Trade Practices Act. Uthmeier said penalties could total billions of dollars.

In other words: pay for the past harms and change the product going forward.

In a statement to The Conversation, an OpenAI spokesperson pointed to the company’s “industry leading protections and policies” regarding user safety.

In particular we built safety for minors directly into our products, including a more protective experience specifically for minors, an age prediction tool, defaulting users whose age we are not confident into our more protective experience, and giving parents tools to monitor their kids use of AI.

Adding to a growing pile

This lawsuit is a significant development, but it has not arrived in a vacuum.

Across the US, the courts are filling with cases accusing tech companies of harming young people. In April, for example, Uthmeier launched a criminal investigation into OpenAI over the chatbot’s alleged role in a shooting at Florida State University.

Some juries have started to side with the plaintiffs.

In March 2026, for example, a New Mexico jury hit Meta with a US$375 million penalty in a child safety case. Days later, a jury in Los Angeles found Meta and Google liable in a landmark trial over social media addiction.

This case rides the same current. But it broadens the scope by alleging Altman himself should be personally responsible.

Uthmeier is demanding a trial by jury.

Alexandra Andhov is the director of ALTeR (Center for Advancing Law and Technology Responsibly) at the University of Auckland. She received funding from the Independent Research Fund Denmark for the "PROFIT" Project (Gaps and Opportunities in Corporate Governance of Big Tech Companies) to research big tech companies.

This week’s magnitude 7.8 earthquake in the Philippines came with scenes familiar to New Zealanders: collapsed buildings, shattered facades and streets strewn with rubble.

Earthquakes of such force test buildings to their extreme limits. As occurred this week, and in Christchurch in 2011, some ultimately fail with tragic consequences.

But, for structural engineers, preventing collapse is only part of the challenge. Increasingly, we are also asking what happens to buildings that survive major earthquakes.

Many modern buildings are designed to protect lives, but often at the cost of damage that can take years and millions of dollars to repair. Some quake-damaged buildings have ultimately been demolished despite never having come close to collapse.

On top of this, the construction sector is under growing pressure to shrink its substantial share of global greenhouse emissions. This is raising the need for building systems that are sustainable and resilient.

Last month, in one of the country’s most demanding full-scale earthquake tests, we assessed an emerging timber-based technology to find that it can meet all these requirements.

Buildings that bounce back

Over the past decade, many people will have heard growing talk about timber as a low-carbon alternative to concrete and steel. While we might picture traditional timber-framed houses, modern mass timber construction is very different.

One of the most promising products – cross-laminated timber (CLT) – is made by bonding layers of timber boards together at right angles, creating large structural panels that can be used to construct multi-storey buildings.

As a renewable material, it stores carbon absorbed during tree growth and can reduce the embodied emissions of buildings compared with concrete and steel. It is also well suited to prefabrication, with entire building components or modules manufactured off-site and assembled later, reducing construction time, waste and disruption.

During earthquake shaking, engineered timber structures have been found to perform extremely well. But less understood is how these new modular mass timber buildings accommodate movement.

To model this, we developed a system that allows storeys to move relative to one another during an earthquake, rather than forcing the entire building to act as a rigid unit.

This controlled movement reduces the strain placed on the building during an earthquake. Once the shaking stops, the system helps the structure return to its original position, reducing damage and improving the chances it can be used again quickly.

Putting timber to the test

To understand how our system performs under realistic earthquake conditions, we built a full-scale, modular CLT building and tested it on the University of Auckland’s “shake table” simulator.

While the test building was physically two storeys high, additional weight was added at roof level to replicate the forces experienced by a typical three-storey building – one of the most common forms of medium-density housing in New Zealand.

The simulation itself subjected the building to a series of increasingly demanding earthquake shaking, reflecting what would be experienced from both distant and near-source events.

The building performed as hoped, with the connection system allowing each storey to move in a controlled way during the simulated earthquakes. This helped absorb and dissipate energy while protecting the main timber structure from damage.

Perhaps most importantly, the building returned to its original position after the shaking stopped, rather than being left permanently tilted or displaced. Engineers call this “self-centring” – a key feature of buildings designed not just to survive earthquakes, but to recover from them.

And while the building moved about during the shaking, the main timber structure remained undamaged. In a real-world earthquake, that could mean lower repair costs, less disruption and a faster return to normal use.

There were, however, questions our test could not answer. For instance, it did not assess how non-structural elements such as wall linings, services and interior finishes – which are often damaged during earthquakes – would perform in a real building.

Nevertheless, the results provide encouraging evidence that modular timber buildings can be designed not only to withstand major earthquakes, but also to recover from them with minimal damage.

The next step is to incorporate the technology into complete building systems and assess its long-term performance, practicality and commercial viability.

If those hurdles can be overcome, it could help support a new generation of low-carbon buildings that are safer, more resilient and quicker to return to service after major earthquakes.

As countries such as New Zealand continue to grapple with both seismic risk and the need to reduce construction emissions, innovations like these may help show that resilience and sustainability do not have to come at the expense of one another.

Ashkan Hashemi receives funding from Natural Hazard Commission Toka Tū Ake.

The US state of Florida has filed a lawsuit against OpenAI and Sam Altman, alleging the tech giant and its CEO put profit over public safety with its flagship artificial intelligence (AI) product, ChatGPT.

The lawsuit, filed in Florida state court on Monday local time by Florida’s attorney general James Uthmeier, is one of the most significant enforcement actions brought by a state attorney against an AI company to date.

It comes as OpenAI and other big tech companies are embroiled in a growing number of legal cases related to the alleged harm their products have caused.

Six key elements

The complaint opens with a screenshot of OpenAI’s own parental-control page, which states that ChatGPT was “built with safety in mind”. Then, in a standalone paragraph, the State answers with two words: “Not so”.

This signals the central allegation of the case: that OpenAI sold ChatGPT to the public as safe and reliable, while knowing it could cause serious harm

More specifically, there are six key themes to Florida’s case against OpenAI. The first is that the company engaged in deceptive safety marketing, assuring parents the platform is safe for teenage use, while not clearly disclosing that ChatGPT can be wrong.

Second, despite OpenAI’s marketing, ChatGPT is unreliable. A 2025 study, for example, found AI assistants, such as ChatGPT, misrepresent the news roughly 45% of the time. Similarly, despite marketing suggesting ChatGPT can handle financial affairs, ChatGPT has failed in meeting basic accounting standards and provided incorrect tax advice to users.

The third element of the case is the public safety threat. The danger to young people in particular is illustrated by the tragic story of Adam Raine, a 16-year-old who died by suicide in April 2025 after engaging in long conversations with ChatGPT. When Adam expressed suicidal thoughts, ChatGPT responded that it “won’t try to talk you out of your feelings”. It helped Adam plan a “beautiful suicide” and even offered to write his suicide note for him.

Why would a product behave this way? Because, Uthmeier argues, it was built to.

OpenAI designed ChatGPT to be highly agreeable, to say “yes” roughly ten times as often as “no”, according to a Washington Post review of 47,000 conversations. This forms the fourth element of the case – commercial exploitation through sycophancy. In other words, ChatGPT optimistically parrots back users’ responses in order to to manipulate them into deeper conversations, regardless of truth or safety.

But according to the lawsuit, even ordinary use carries a cost: it weakens people’s brain activity and critical thinking skills (also known as cognitive atrophy). This is the fifth element of the case.

The sixth and final element is knowledge – specifically, the knowledge of Samuel Altman. According to Uthmeier, since at least 2023, OpenAI’s own documents warned that the model could coach people on committing crimes, but Altman overruled the safety staff.

These six elements paint a picture of a product marketed as safe, engineered to be addictive, and known by its own makers to be dangerous – yet sold to us, anyway.

Altman is at the centre of that picture. The complaint reconstructs his career and reaches for an April 2026 New Yorker investigation and testimony from the recent legal battle between Elon Musk and OpenAI to depict a man, who in Uthmeier’s telling, repeatedly chose speed over safety.

That is why Uthmeier is asking the court to hold Altman personally liable for “his utter disregard for the risk to human life”.

Pay for past harms

Uthmeier is asking the court to declare that OpenAI broke the law, then to order the company to stop – permanently – its unlawful practices.

He wants the company barred from collecting children’s data without parental consent and the safeguards that should come with it, and barred from misrepresenting or staying silent about ChatGPT’s risks.

On top of the injunctions, the state is seeking civil penalties of up to US$10,000 per violation for OpenAI’s alleged wilful violation of the the Florida Deceptive and Unfair Trade Practices Act. Uthmeier said penalties could total billions of dollars.

In other words: pay for the past harms and change the product going forward.

In a statement to The Conversation, an OpenAI spokesperson pointed to the company’s “industry leading protections and policies” regarding user safety.

In particular we built safety for minors directly into our products, including a more protective experience specifically for minors, an age prediction tool, defaulting users whose age we are not confident into our more protective experience, and giving parents tools to monitor their kids use of AI.

Adding to a growing pile

This lawsuit is a significant development, but it has not arrived in a vacuum.

Across the US, the courts are filling with cases accusing tech companies of harming young people. In April, for example, Uthmeier launched a criminal investigation into OpenAI over the chatbot’s alleged role in a shooting at Florida State University.

Some juries have started to side with the plaintiffs.

In March 2026, for example, a New Mexico jury hit Meta with a US$375 million penalty in a child safety case. Days later, a jury in Los Angeles found Meta and Google liable in a landmark trial over social media addiction.

This case rides the same current. But it broadens the scope by alleging Altman himself should be personally responsible.

Uthmeier is demanding a trial by jury.

Alexandra Andhov is the director of ALTeR (Center for Advancing Law and Technology Responsibly) at the University of Auckland. She received funding from the Independent Research Fund Denmark for the "PROFIT" Project (Gaps and Opportunities in Corporate Governance of Big Tech Companies) to research big tech companies.

On Monday, a nine-member federal jury in Oakland, California took less than two hours to dismiss Elon Musk’s lawsuit against OpenAI and its chief executive Sam Altman.

Crucially, the jury did not rule on the core claims of the case. These included whether OpenAI, the company behind the popular artificial intelligence (AI) chatbot ChatGPT, strayed from its founding mission and whether Altman and OpenAI’s co-founder Greg Brockman enriched themselves at the expense of a charitable purpose.

It decided only that Musk had waited too long to sue in relation to his core claims about breaches of a founding contract or breach of charitable trust.

A victory for Musk could have neutered OpenAI, which in turn would have probably sent shockwaves through the entire AI sector given the company’s dominant position developing the technology.

Now, however, OpenAI has a clear path to take its next big step in the AI race, even though the key question at the core of the case remains unanswered: is OpenAI a nonprofit dedicated to humanity or a corporation dedicated to its shareholders?

How it all started

OpenAI was founded in December 2015 as a nonprofit entity – an AI research lab.

Musk and a group of prominent entrepreneurs pledged US$1 billion to develop AI for the benefit of humanity, free of commercial pressure. Alongside Musk, the founding group included Altman, Brockman and computer scientist Ilya Sutskever.

The organisation’s charter committed to two key principles. First, developing artificial general intelligence safely and for the benefit of all of humanity.

Second, developing the technology openly, meaning it would be open source. This would allow others to use their underlying models, code, and research freely.

This was the deal Musk says he signed up for. And OpenAI claims it continues to honour this deal even today, despite more than US$20 billion in revenue in 2025.

Since 2015, a lot has happened. And understanding these events is key to interpreting the jury’s verdict.

A very different deal

By 2019, the original deal looked different. Given that training frontier AI models was extraordinarily expensive, Altman started to seek more cash.

OpenAI created a capped-profit subsidiary where investors could earn up to 100 times their initial investment, with any extra money flowing back to the nonprofit parent.

One of the first investors was Microsoft, which initially invested US$1 billion and more than US$13 billion over time. The nonprofit retained formal governance, the usual nonprofit rules applied, but the commercial subsidiary became the decision-maker.

That same year, OpenAI released GPT-2. The model was released partially, in stages, rather than published as open source. This was the moment the “open” in OpenAI began to read differently.

GPT-3 followed in 2020, and it was available only via a paid subscription. The inner workings of the model also remained secret. ChatGPT launched in November 2022, and reached 100 million users in a few days.

Twelve months later, OpenAI’s nonprofit board fired Sam Altman, citing a loss of confidence in his candour. This was what the governance structure was meant for: to protect the organisation’s humanity-first mission, the board had the power to remove the chief executive.

Yet, within five days, after pressure from Microsoft and the employees, Altman was back and the board was out. A new board that aligned with the commercially-driven enterprise took their seats.

The mechanism built to keep OpenAI accountable to its charter was the one that lost. Whatever the “humanity claim” of the founding mission was supposed to mean, commercial interests prevailed.

A sweeping reorganisation

In October 2025, after nearly a year of negotiation with the attorneys general of California (where OpenAI is headquartered) and Delaware (where it is incorporated), the organisation completed a sweeping reorganisation.

The nonprofit became the OpenAI Foundation, with the same mission: “to ensure artificial general intelligence benefits all of humanity”. The for-profit became a public benefit corporation, called OpenAI Group PBC. Unlike a conventional corporation, it is required to advance its stated mission and consider the broader interests of all stakeholders.

The OpenAI Foundation holds a 26% stake in the new public benefit corporation and retains some contractual and special shareholder governance rights. Microsoft owns 27% and the remaining 47% is owned by other investors and employees.

Thus the Foundation controls the public benefit corporation in form. Yet in practice, OpenAI is now a profit-seeking enterprise with a charitable shareholder. So while a number of nonprofit governance guardrails are in place, significant deficiencies remain.

The unanswered question

OpenAI is now openly preparing for a public listing at the end of 2026, at an expected valuation at up to US$1 trillion, even as it defends dozens of pending lawsuits, ranging from intellectual property infringement and consumer protection claims to a wrongful death suit.

This is the part the jury did not address.

A verdict on a statute of limitations is a statement about timing, not purpose. It tells us when a complaint can be heard. It does not tell us whether the complaint was right. And in this particular case, it demonstrates the difficulty in relying on private individuals to enforce non-profit governance norms.

Musk has said he will appeal the verdict. The appeal court will almost certainly limit itself to a narrow legal question – perhaps when a reasonable plaintiff should have understood OpenAI had changed.

The larger question about whether OpenAI is a nonprofit dedicated to humanity or a corporation dedicated to its shareholders, has now been deferred indefinitely – at least in a legal context.

The public, however, will no doubt make up its own mind about a company now worth hundreds of billions of dollars.

Alexandra Andhov is the director of ALTeR (Center for Advancing Law and Technology Responsibly) at the University of Auckland. She received funding from the Independent Research Fund Denmark for the "PROFIT" Project (Gaps and Opportunities in Corporate Governance of Big Tech Companies) to research big tech companies.

Ian Murray is a Professor of Law at the University of Western Australia, Director of the Charity Law Association of Australia and New Zealand and also a member of the Law Council of Australia’s Charity and Not-for-profits Sub-committee.