The Lancet has flagged nuclear conflict and the malicious application of AI as two critical risks of human extinction, but their regulatory landscapes appear to be poles apart. I believe our approach to preventing nuclear conflict is far more comprehensive than the measures we have adopted to regulate AI, and I base my view on three factors: legal obligations, the operational authority of regulators, and readiness for implementation.

In the nuclear field, the Treaty on the Non-Proliferation of Nuclear Weapons (191 states parties) and the Treaty on the Prohibition of Nuclear Weapons (adopted by 122 votes) create legally binding duties for their respective parties. IAEA safeguards are backed up by an inspection regime, and the Preparatory Commission for the Comprehensive Nuclear-Test-Ban Treaty Organization operates a global monitoring network, with 307 certified facilities reported in February 2026.

In stark contrast, the Joint Committee on Human Rights of the UK Parliament has warned that no country currently has an adequate approach to AI legislation and regulation. Existing legislation is fragmented, and the interactions between AI safety institutions and model builders can lack binding legal authority. Although the EU AI Act is already being implemented, some core obligations for high-risk AI systems will only apply from 2027 and 2028 respectively. According to OECD data, among 36 surveyed countries, only 12 reported independent oversight for government use of generative AI, while seven reported accountability frameworks. Nuclear regulation has established verification mechanisms, while comparable global AI governance remains fragmented.

The nuclear regime is, of course, far from perfect. New START expired in February 2026, with no successor treaty in place, while nuclear modernisation continues.

There has also been some movement in AI regulation. California’s Transparency in Frontier Artificial Intelligence Act (SB 53), effective in 2026, establishes transparency, safety reporting and accountability requirements for major frontier AI developers. However, these efforts remain relatively new.

None of that changes my conclusion: without a global, institutionally credible, verifiable and enforceable regime comparable to established nuclear governance mechanisms, AI regulation remains fragmented across national and regional jurisdictions.

The seriousness of regulation must match the level of threat. If both risks have extinction potential, humanity should give AI regulation stronger legal obligations, effective oversight agencies and enforceable safety requirements.

In October 2026, 《The Lancet》 released the “Report of the Commission on Global Health Threats in the 21st Century.” This report highlights two primary risks that could endanger human survival or potentially result in human extinction—before 2100: the malicious use of artificial intelligence and nuclear conflict, including secondary effects like famine caused by nuclear events. The report also describes 15 interconnected catastrophic health threats, covering areas like climate change, infectious diseases, socioeconomic challenges, and environmental hazards. Using predictive models and Disability-Adjusted Life Years (DALYs) as a measurement, the study evaluates the possible effects of these threats on global population health over the next 75 years.

Various media have handled the coverage of the study differently. This comparison looks at the handling of the study across a number of parameters: headlines and sub-headings content and structure data, evidence, and coverage wider picture.

Headline-wise, The Independent and Sky News used identical headlines that put a specific emphasis on “human extinction” as a potentiality. The Guardian opted to zoom in on the “malicious use of AI” as the cause, while Euronews wrote that it “could end humanity” and included the phrase “scientists warn” to reinforce the potential and the scientific warning behind it.

Where as in this case on Instagram it is more free form, with the use of the inquisitive “Could Humanity Face Extinction?”, drawing the reader in and increasing the interactivity. This marks the difference in how information is put across in the news media and social media.

Second is subheading The Independent gives its numbers to show “the scale of potential losses that can be averted” while The Guardian and Euronews adopt a more blunt positioning in a strong “call for action”. Sky News focuses on “the full extent of the threat”.

While all four outlets derived from the same Lancet report, their coverage was different in scope. While the Independent presented the existential risk from AI and nuclear war, it also provided a broader overview of global health trends and how they affect global politics. Sky News prioritized government action and the need for urgent interventions. The Guardian put a greater importance on global health inequalities, and how international aid cuts are affecting poor nations. Euronews dealt more with scientific data and technological solutions, while Instagram summarized the main risks, presenting a lot of eye-catching visuals that provided the main information to readers and additional personal commentary on government regulation for some posts.

Even though all four media outlets relied on the same Lancet report for their coverage, each emphasized different angles. The Independent wrote about “the public health picture and changes to international politics”. Sky News concentrated on “decision making within the UK government and the measures in place to respond”. The Guardian focused on “global inequalities of health” and Euronews built a lot of the report around “scientific data and technological solutions”. Instagram but attracts engagement through being easy to digest and via more personal perspectives. In comparison, mainstream news outlets presented “more comprehensive coverage, accompanied by context and analysis from many perspectives”.