A post has been circulating with a burning globe and a line in block capitals: Earth is on track to become uninhabitable and cross a climate “point of no return,” scientists say. Underneath it, a second claim: warming is rising exponentially, and far faster than anyone predicted.
There is a real paper behind this. It is serious, it is recent, and it is written by people whose names carry weight. It also says none of those three things.
What the paper actually says
The citation is Ripple, Wolf, Rockström, Richardson, Wunderling, Gregg, Westerhold and Schellnhuber, “The risk of a hothouse Earth trajectory,” One Earth 9(2), article 101565. The first thing worth knowing is its category. It is a commentary — a short, peer-reviewed argument piece that synthesises existing literature. It is not a new model run, a new dataset, or a new observation. That does not make it worthless; commentaries by researchers of this seniority are how a field signals that it thinks the risk picture has shifted. It does mean the paper produces no new numbers of its own.
Its abstract, in full, reads: “Earth’s climate is now departing from the stable conditions that supported human civilization for millennia. Crossing critical temperature thresholds may trigger self-reinforcing feedbacks and tipping dynamics that amplify warming and destabilize distant Earth system components. Uncertain tipping thresholds make precaution essential, as crossing them could commit the planet to a hothouse trajectory with long-lasting and potentially irreversible consequences.”
Count the hedges: may trigger, could commit, potentially irreversible, uncertain thresholds. That is not a forecast. It is an argument that the uncertainty itself is the reason to act, because you cannot un-cross a threshold once you find out where it was.
The authors are explicit that they are describing a risk, not a schedule. Rockström put it plainly to Yale Environment 360: “Our paper shows that we’re not there yet — but we’re very close.” Co-author Christopher Wolf: crossing even some thresholds “could commit the planet to a hothouse trajectory with long-lasting and possibly irreversible consequences.” The strongest scenario language in any of the associated press material is “extreme warming and sea level rise — conditions that could be difficult to reverse on human timescales, even with deep emissions cuts.”
“Difficult to reverse” and “uninhabitable” are not the same sentence.
The three viral claims, checked
1. “Uninhabitable” is a headline, not a finding
The word traces to a syndicated aggregation piece, not to the researchers. It is worth being precise about what climate scientists actually say on this, because the honest version is bad enough without inflation.
Whole-planet uninhabitability this century is not a scenario anyone credible holds. MIT’s Adam Schlosser puts the odds of climate change driving human extinction at “very low, if not zero,” and notes that the IPCC lists many severe risks but not that one. The physiological work is even clearer on scale: Sherwood and Huber’s 2010 PNAS analysis found that roughly 7°C of global warming would make some regions physiologically uninhabitable, and 11–12°C would extend that to most of where people currently live. Those are numbers far outside any plausible 2100 pathway.
What is real is regional. Raymond, Matthews and Horton reported in Science Advances in 2020 that a handful of coastal stations in the Persian Gulf and the Indus Valley have already briefly touched the 35°C wet-bulb limit — the point at which a healthy person in the shade with unlimited water can no longer shed heat — and that extreme humid heat has more than doubled in frequency since 1979. Those events lasted one to two hours. That is the defensible version of the habitability story: not a dead planet, but a growing number of places where being outdoors at the wrong hour can kill you.
2. “Exponential” is the wrong word — and the right one is still alarming
No major climate agency describes global temperature rise as exponential. Not the IPCC, not NOAA, not Copernicus, not Berkeley Earth. The functional forms researchers actually fit are linear trends, piecewise-linear trends with a change point, and quadratics.
There is a physical reason. Radiative forcing scales with the logarithm of CO₂ concentration, and CO₂ concentration is rising roughly linearly. A logarithm of a near-linear input cannot produce exponential output. When statisticians fit a free exponent to the temperature record, they find it only rises above 1 after about 2009 and stays below 2 — supra-linear, sub-quadratic. Exponential growth is far steeper than quadratic. The data rule it out.
But “accelerating” is a different claim, and that one now has real support. Foster and Rahmstorf published an analysis in Geophysical Research Letters in 2026 that removes the El Niño, volcanic and solar signals from five independent temperature datasets and finds a statistically significant acceleration, with a change point around early 2013. The Indicators of Global Climate Change 2025 assessment, published in June 2026, puts the current rate of human-induced warming at 0.27°C per decade for 2016–2025, against roughly 0.18–0.20°C per decade for the late twentieth century.
Even this is contested inside the field, and the disagreement is worth knowing. The acceleration is not visible in the raw temperature series — only after natural variability is subtracted, and that subtraction is, in Zeke Hausfather’s words, “more of an art than a science.” Piers Forster, who leads the annual indicators update, calls “accelerating” a word that is “emotive, but difficult mathematically.” Michael Mann attributes most of the 2023–24 spike to El Niño plus the unmasking effect of cleaner shipping fuel rather than a steepening underlying trend.
3. “Faster than predicted” is close to the opposite of true
This is the claim that collapses most completely. Hausfather and colleagues evaluated seventeen climate model projections published between 1970 and 2007 against what actually happened, Ten of the seventeen matched observed temperature directly; fourteen matched once you correct for the fact that the emissions those models were fed differed from the emissions the world actually produced. Their conclusion: “no evidence that the climate models… have systematically overestimated or underestimated warming.”
The acceleration itself was predicted. IPCC AR6 projects 0.24°C per decade for 2015–2050 and CMIP6 projects 0.29°C per decade — both faster than the observed post-1970 rate. As Gavin Schmidt wrote in 2024, “everybody is (or should be) expecting an acceleration.” Observing one is not evidence that the science was wrong. It is the science being right.
What a tipping element actually is
A tipping element is a part of the climate system with a self-reinforcing feedback strong enough that, past some threshold, it keeps changing on its own even if you stop pushing. The canonical modern assessment is Armstrong McKay and colleagues in Science in 2022, which put numbers on sixteen of them.
Two things about that table get misread constantly. First, the thresholds are estimates with enormous ranges, not measurements. Second, the timescales attached to them describe how long the transition takes once triggered, not how long until it triggers. Greenland’s “10,000 years” means the collapse plays out over millennia — not that nothing happens for ten thousand years.
The thresholds keep moving — and they keep moving down
This is the part of the story that actually justifies a new commentary in 2026, and it is the part the viral post never mentions. Since the 2022 assessment, the Global Tipping Points Report published in October 2025 — 160 scientists across 87 institutions, led from the University of Exeter — revised several central estimates, and almost every revision went the same direction.
Coral’s central threshold moved from 1.5°C down to 1.2°C. The Amazon’s lower bound moved from 2.0°C down to 1.5°C. The report puts the ice sheets at risk “since at least 1°C of global warming,” and a February 2026 Nature Climate Change paper by Winkelmann, Garbe and colleagues found a first Antarctic threshold “potentially as low as 1–2°C above pre-industrial levels” — enough to commit roughly 40% of West Antarctica’s marine ice volume to long-term collapse. And where the 2022 table gave AMOC a central estimate of 4°C, the 2025 report simply says tipping “cannot be ruled out at current warming levels.”
That is the real news: not that a date has been set, but that the error bars have shifted toward us.
The cascade is the actual argument
The commentary’s central claim is not about any single element. It is that the elements are coupled, and that coupling is mostly destabilising. In the 2025 report’s assessment of twenty climate tipping-system interactions, most amplify rather than damp — and AMOC alone features in 45% of them.
One threshold has already been crossed
If you want the single most defensible alarming fact in this whole story, it is not about ice or the Amazon. It is coral.
The fourth global bleaching event ran from early 2023 to mid-2025 and exposed 84% of the world’s reef area, across at least 83 countries and territories, to bleaching-level heat stress. It is the most severe on record. On that basis the 2025 report concluded that warm-water coral reefs have passed their thermal tipping point — the first Earth system judged to have done so — and described them as “virtually certain” to tip.
That is a threshold crossed at roughly 1.2°C, in a world that has already warmed more than that. It is also the clearest illustration of what these transitions actually look like from inside: not a bang, but a bad year that does not recover, followed by another.
AMOC: the most misrepresented number in climate science
The Atlantic Meridional Overturning Circulation is the ocean conveyor that keeps northwest Europe unusually mild. It is also where climate coverage goes furthest off the rails, because in 2023 Peter and Susanne Ditlevsen published an estimate that it could collapse around 2057, with a confidence interval of 2025 to 2095. That number went everywhere.
What travelled less well was everything around it. The estimate rests on early-warning statistics applied to a sea-surface-temperature proxy, and assumes a particular mathematical form of collapse; the authors themselves wrote that “we cannot rule out that other mechanisms are at play, and thus, the uncertainty is larger.” A separate 2024 analysis by Maya Ben-Yami, Niklas Boers and colleagues at the Technical University of Munich and the Potsdam Institute — tipping-point specialists, not skeptics — found that the uncertainties permit collapse dates anywhere from 2050 to 8065.
A 2025 Nature study led by Jonathan Baker concluded the AMOC is “unlikely to collapse this century” while remaining very likely to weaken. Stefan Rahmstorf, who takes the risk more seriously than most, makes a risk-management argument rather than a forecast: even a hypothetical 40% chance that the Ditlevsen estimate is right, he has written, would be a major change from the IPCC’s assessment that the risk is under 10%. On the underlying slowdown, he and Levke Caesar wrote in a 2026 preprint that it “may not be certain, but it is at least very likely.” Nobody serious is claiming a date.
Climate scientists are arguing about the framing too
It would be dishonest to present tipping points as settled. There is a real, good-faith split inside the field, and it is not between believers and deniers — everyone in it agrees emissions cuts are urgent.
On one side, Paul Valdes has argued since 2011 that climate models are “built for stability” and may systematically understate the potential for abrupt change, because they struggle to reproduce abrupt events in the paleoclimate record without unrealistic forcing. The absence of tipping in a model is not evidence of absence.
On the other, Robert Kopp and colleagues published a critique in Nature Climate Change in 2024 critiquing the framing “for oversimplifying the diverse dynamics of complex natural and human systems and for conveying urgency without fostering a meaningful basis for climate action.” Kopp’s follow-up put it more bluntly: tipping points work “better as a metaphor than as describing a technically defined class of processes.” A 2026 paper in Climatic Change warned that “perceived overstatement in this context may weaken support for climate policy in general.”
Michael Mann has spent years making exactly this point: “Exaggeration of the climate threat by purveyors of doom — we’ll call them ‘doomists’ — is unhelpful at best. Indeed, doomism today arguably poses a greater threat to climate action than outright denial. For if catastrophic warming of the planet were truly inevitable and there were no agency on our part in averting it, why should we do anything?”
What the authors actually recommend
The commentary’s prescriptions are unglamorous and specific: scale up renewables, protect carbon-storing ecosystems, embed climate resilience into government policy, phase out fossil fuels in a way that is socially just, build a coordinated global monitoring network for tipping points, and improve systemic risk management.
That last one is the most interesting and the least covered. Right now nobody is systematically watching these systems for early-warning signals in a coordinated way — which is precisely why the timing estimates are so wide. And the authors are careful to end on agency rather than fatalism: preventing even a tenth of a degree of warming, they argue, measurably reduces the odds of crossing a threshold. That only makes sense if the outcome is still open.
Frequently asked questions
Is the “hothouse Earth” study real?
Yes. Ripple et al., “The risk of a hothouse Earth trajectory,” One Earth 9(2), article 101565, published 11 February 2026. It is a peer-reviewed commentary rather than original research, written by eight scientists from Oregon State, the Potsdam Institute, IIASA, the University of Copenhagen and MARUM.
Does it say Earth will become uninhabitable?
No. That word appears in an aggregated headline, not in the paper, its abstract, or any of the institutional press releases. The paper’s own worst-case language is “extreme warming and sea level rise” that could be “difficult to reverse on human timescales.”
What temperature is “hothouse Earth”?
The term comes from a 2018 PNAS paper by Steffen and colleagues, which described an end state around 4–5°C above preindustrial with sea level 10 to 60 metres higher than today — on a multi-century to millennial timescale, not by 2100. That caveat is almost always dropped in coverage. The 2026 commentary does not derive a temperature of its own.
Have we passed a point of no return?
For warm-water coral reefs, the 2025 Global Tipping Points Report concludes yes, at roughly 1.2°C. For everything else, no — and the honest position is that nobody can currently say where the other thresholds are, only that the estimates have been moving downward.
Is global warming accelerating?
Probably yes. The current rate of human-induced warming is about 0.27°C per decade against roughly 0.18–0.20°C in the late twentieth century. But the acceleration is only statistically visible once El Niño and volcanic effects are stripped out, several senior researchers dispute the magnitude, and the acceleration was itself projected by IPCC AR6. It is not exponential.
Was 2025 the hottest year on record?
No — 2024 was, at about 1.55°C above preindustrial. 2025 came in at 1.44 ± 0.13°C, ranking second or third depending on the dataset, despite starting and ending in a cooling La Niña. The 2023–2025 three-year average came in at 1.48 ± 0.13°C — close to the Paris threshold, but not past it.
Sources
Ripple et al., One Earth 9(2):101565 (2026) · Armstrong McKay et al., Science 377:eabn7950 (2022) · Global Tipping Points Report 2025, University of Exeter · Forster et al., Earth System Science Data 18:3889 (2026) · Foster & Rahmstorf, Geophysical Research Letters (2026) · Hausfather et al., GRL 47 (2020) · Ditlevsen & Ditlevsen, Nature Communications 14:4254 (2023) · Ben-Yami et al., Science Advances (2024) · Kopp et al., Nature Climate Change (2024) · Steffen et al., PNAS 115:8252 (2018) · Raymond et al., Science Advances 6 (2020) · IPCC AR6 WG1 · WMO, Copernicus, Berkeley Earth, NOAA GML, Scripps CO₂ Program.
