Sustainability update September 2026

UN report details “best remaining option” to limit global climate overshoot

Published in September, the UN Environment Programme’s Limiting Overshoot climate report examined the risks and impacts of higher peak temperature “overshoot” beyond the Paris Agreement’s 1.5°C warming target and mapped out mitigation pathways for governments and societies to limit periods of peak exceedance and reduce climate-related vulnerabilities. The report warned that irreversible climate tipping points would likely be reached as the magnitude and duration of higher temperatures increased over time – even the most optimistic predictive peak-temperature models forecast that future warming could rise to 1.8°C above pre-industrial levels.

While continuing to prioritise conventional mitigation and the accelerated adoption of clean energy, the report recommended that governments and social institutions aim for an “overshoot, peak and decline” climate mitigation pathway while also adapting to unavoidable impacts. The pathway follows a trajectory from sustained warming above 1.5°C and a peak below 2°C to mitigation- and adaptation-driven temperature decline and eventual stabilisation. UNEP stressed that, with no good outcomes above 1.5°C, the world should focus on shortening the duration of sustained warming as much as possible and stabilising below the objective threshold as quickly as possible. Here, net zero efforts should be treated as a milestone rather than an endpoint. Progressing from overshoot to decline will require net-negative emissions, removing more CO₂ from the atmosphere than is currently emitted. Removal strategies should be scaled up but mustn’t be used as a “silver bullet” substitute for essential emissions cuts – UNEP also recognised that these strategies can be compromised by ecological, technological and governance constraints. 

Climate adaptation must strengthen in tandem with mitigation, as even a successful stabilisation below 1.5°C would not reverse all the impacts of a sustained peak-temperature overshoot. Governments seeking to reduce emissions must simultaneously work to reduce vulnerabilities associated with climate risks such as extreme weather, excess heat, rising sea levels, and ecosystem failure. Countries and communities that are most vulnerable to the impacts of climate change are often among those that have contributed least to its causes, emphasising the need for equity and justice considerations to be at the heart of all mitigation and adaptation strategies.

UNEP’s report concluded that the world is unlikely to return to “normal” when the temperature decline and stabilisation phases are reached. Surpassing the 1.5°C threshold will result in unavoidable changes to ecological and societal systems, the scale and ramifications of which are largely unknown. With future consequences difficult to predict, UNEP positioned its mitigation pathway as a “crucial scientific frontier” to inspire new research into how the planet might respond to falling temperatures following a peak.

Survey of chief sustainability officers (CSOs) shows that most see climate progress driven by global commerce and technology

An inaugural survey of CSOs by the World Economic Forum (WEF) found that despite escalating conflict and associated geoeconomic volatility, the transition towards global climate goals was showing clear signs of progress driven by growing commercial and technological momentum. Based on a survey of 103 CSOs across five continents, the WEF’s Chief Sustainability Officers’ Outlook report assessed the current level of global sustainability leadership sentiment and detailed how companies were navigating the sustainability transition within a changing economic, political and technological landscape.

Over two-thirds of CSOs surveyed expected global sustainability progress to hold steady or accelerate over the next year, while three-quarters expected companies’ transition-related investment to do the same, driven by a clear economic imperative and increasingly applicable technologies. Stronger business cases and the proliferation of cheaper technologies also featured as the top transition accelerators for the next three years. Sustainability is moving deeper into commercial, operational and investment decisions, and most CSOs emphasised execution over ambition as assets and supply chains are locked into long-term sustainability impacts. AI was expected to significantly accelerate short-term sustainability progress in measurement, reporting, efficiency and risk modelling, though many CSOs acknowledged that the energy and resource intensity of AI infrastructure counterbalanced the more positive impacts. 

Climate adaptation is increasingly viewed as essential for business continuity and resilience. Around 85% of respondents expected adaptation to become a greater global priority over the next three years, with most agreeing that upscaling would rely on greater private-sector investment. Companies already investing in resilience strategies have seen substantial benefits: in 2024, CDP estimated that businesses proactively reducing Scope 3 emissions had saved $13.6 billion in costs, with almost $165 billion in untapped financial benefits still available. However, 62% of CSOs identified ongoing uncertainty around the costs and benefits of adaptation technologies as a principal barrier to increased investment.  

“85% of respondents expected climate adaptation to become a greater global
priority over the next three years.”

Prominent headwinds related to conflict, inconsistent policy and short-term business pressures persist, but CSOs agreed that these will result in a climate transition “green divergence” rather than a significant loss of momentum. Priorities and the pace of change will increasingly diverge across sectors and regions, and markets will respond differently to economic imperatives, policy direction and board-level decision-making. Climate progress will continue in a global green economy predicted to surpass $7 trillion by 2030. 

Distinguishing between facts and hypotheses in the AI risk debate

For years, scientists, security analysts and AI developers have warned of the harmful and potentially existential impacts of advanced AI. Artificial intelligence creates a broad range of risks, from observable threats in deployed systems to speculative outcomes involving future systems with capabilities that are yet to be fully explored or demonstrated. Understanding the distinction between evidenced and hypothetical risks is essential for making sound decisions about AI safety, governance, investment and public policy. Across the developmental landscape, there are risks supported by evidence and observation or established scientific mechanisms, risks that depend on uncertain future development trajectories, and risks that require several speculative assumptions to converge and hold.

Recent incidents show that AI-related risks are no longer purely theoretical. In July, reports suggested that autonomous AI agents developed using OpenAI technology bypassed testing controls and carried out unauthorised cyber activity. AI platform Hugging Face reported a breach in which an AI-driven attack exploited weaknesses in its data processing systems, gaining access to internal datasets and service credentials. This was considered one of the first publicly reported cases of an autonomous AI agent successfully compromising a company from end-to-end. In September, it was reported that an autonomous AI agent had accessed a government Medicare statistics portal in Australia, obtaining non-public information and implanting new files.

AI-enabled cyberattacks have already caused significant financial and operational harm. Examples include the theft of 37 million customer records in a T-Mobile breach and a deepfake fraud that reportedly cost engineering firm Arup Group $25 million after attackers impersonated senior executives to authorise a payment.

“AI-enabled cyberattacks have included the theft of 37 million customer
records and a deepfake fraud reportedly costing $25 million.”

However, predictions of rapid, uncontrollable superintelligence or human extinction remain highly speculative. While these scenarios warrant monitoring, and a significant component of our ongoing engagement activity focuses on encouraging responsible AI development and implementation, the evidence available today does not support them as immediate investment concerns.

The focus is shifting away from the plausibility of frontier AI risk scenarios towards assessing whether companies can reliably test AI models for autonomy, deception, cyber capability and a host of other consequential behaviours prior to deployment. Board oversight, security frameworks, incident disclosure, regulatory preparedness and workforce transition planning are also useful indicators in determining a company’s capacity to responsibly deploy and manage increasingly capable AI service models.