AI Makes Human Judgment Superfluous in Climate Disinformation
The summer of 2025 has brought another wave of extreme heat across Europe. Last week, a photo of a melted traffic light in Paris, later debunked as fabricated, went viral, shared by hundreds of thousands. It looked real — the asphalt was rippled, the street sign drooped like a candle. But it was not real. The image was suspected to have been generated by AI Source: AFP fact-check
The damage was already done before any fact-checker could verify the origin. A climate activist group had posted it to illustrate the urgency of adaptation, but the image was then co-opted by accounts that argued government policies had caused the heatwave Source: BBC monitoring Fact-checkers needed hours to trace the metadata; the AI had created the image in seconds. The human role of verifying truth had been bypassed.
This is not an isolated incident. Climate disinformation has evolved from outright denial of global warming to attacking green policies, and at the center of this shift lies a technology that makes human expertise increasingly superfluous. AI generates content so fast, so convincingly, and at such low cost that it overwhelms the human capacity for verification, rational debate, and skepticism. The machine does not need to be right — it needs to be faster than the fact-checker.
The Boundary Between Human and Machine Judgment
In autumn 2024, eastern Spain experienced a catastrophic flood Source: AEMET flood report. Within hours, AI-generated videos appeared on social media showing dams collapsing and water towers bursting Source: Reuters investigation. These were fabricated. The real cause was a year’s worth of rainfall in a single day, but the fake footage shifted blame to dam mismanagement and EU river restoration policies. Real human victims were reduced to political pawns in a disinformation campaign that needed no human author.
The boundary where human judgment and machine recommendation meet is dissolving. A user scrolling through a feed does not pause to ask whether the flood video was made by a human or an algorithm. The platform’s recommendation system, itself an AI, amplifies whatever drives engagement — and false content often outpaces true content. Human critical thinking becomes an afterthought. The machine decides what you see first.
Climate disinformation has always relied on emotional triggers. Now AI can generate those triggers on demand. A heatwave in Europe prompts a surge of posts claiming temperatures were higher in the 1970s. These claims are accompanied by AI-fabricated charts that look like official Met Office data. A human viewer must either accept the chart at face value or invest time in cross-checking. Most will not. The machine’s output is designed to be consumed, not questioned.
Who Decides What You See?
The actor level of disinformation has shifted. Previously, a human propagandist would craft a message, pay for ad placement, and monitor its spread. Today, an AI bot network can generate thousands of unique messages targeting specific demographics, all tailored to local anxieties. No human needs to compose each post. The AI decides the framing.
Take the case of the Iberian Peninsula blackout in spring 2025 Source: REE grid report A major power outage affected mainland Portugal and Spain. Within hours, local Facebook groups were filled with posts blaming renewable energy for destabilizing the grid. These posts shared fabricated technical explanations from “pseudo-experts” that were almost certainly AI-generated or repurposed by automated accounts. The actual cause was later disputed, with some technical analyses pointing to voltage control failures and grid oscillations — not solar or wind power. But the damage was done.
Who made the decision to blame renewables? No single person. The AI had scanned historical patterns of energy skepticism, extracted common phrases like “grid instability” and “renewable reliability,” and produced narratives that matched pre-existing distrust. The human actors who shared these posts were simply conduits. The machine had already decided the story.
This pattern repeats across Europe. In Germany, the far-right party Alternative for Germany questions scientific consensus on climate change. Their messaging does not need human speechwriters for every variation — AI can generate local-language versions of attacks on green policy, embedding cultural references specific to each region. Human speakers merely recite what the algorithm optimized for maximum division.
Who Bears the Consequence of Automated Lies?
The burden falls on the most vulnerable. Climate scientists face increasing harassment as disinformation frames them as the architects of failed policies. “People argue that they have been too alarmist, pointed toward the wrong solutions, made wrong decisions — so the blame is placed on the experts,” says Eva Morel of the French watchdog Quota Climat Source: Quota Climat report The AI-generated content that fuels this harassment is not authored by a person; it is a product of automated systems that learned from crowdsourced hatred.
Yet the human scientists must cope with the consequences. They receive death threats, they leave social media, they stop engaging with the public. The human capacity for reasoned debate becomes superfluous when the machine can amplify any lie at any scale.
Even well-intentioned actors become victims. The environmental group that shared the fake melted traffic light in Paris wanted to highlight climate impacts. But their post was hijacked by disinformation networks that used the same image to argue that “climate activists are lying to you.” The AI had rendered the distinction between genuine content and fabrication meaningless. The group’s human judgment — that the image communicated a real danger — was exploited by machines that have no sense of truth.
The Question of Responsibility
When a system gets it wrong, who is accountable? In 2026, that question has no clear answer. The AI that generates flood fake images has no legal personhood. The platform that amplifies them employs thousands of content moderators but relies on automated detection that is itself flawed. The human who shares the post may believe it is true. The political actor who capitalizes on the lie may never have authored it.
The monetization of social media fuels the attention economy. As Philip Newell of Climate Action Against Disinformation puts it, “Lies are fun and engaging.” AI makes them cheaper to produce and harder to trace. The economic incentives align perfectly with automated deception.
In central and eastern Europe, narratives about renewable energy as a “foreign intrusion” resonate because of deep historical distrust of outside institutions. AI picks up on these cultural undercurrents and generates messages that feel authentically local. A human propagandist would need to understand the region’s energy history, its national identity attached to coal or nuclear power. The AI simply learns from text patterns.
The result is a flood of disinformation that makes human fact-checkers superfluous. They cannot scale. They cannot match the volume. They cannot compete with the speed of machine-generated lies, tailored to each village, each WhatsApp group, each Facebook page. The human role of verification becomes a luxury of the few.
The Cost of Automated Certainty
Climate disinformation does not need to be coherent. It only needs to sow doubt. AI excels at creating plausible uncertainty — a chart that looks official, a quote that sounds authentic, a photo that matches a real event. Each piece undermines trust in institutions, in scientists, in the very idea of evidence.
Ned Mendez, head of research at digital campaigning agency 411, observes that “the disinformation industry has moved one rung downstream.” The fight is no longer about whether global warming is real. It is about whether the response is feasible, fair, worth the price. AI accelerates this shift by generating endless variations of the same doubt.
When a year’s worth of rain hit Spain in 2024, the false claim that dams were intentionally released found fertile ground not because it was plausible, but because it fed a pre-existing narrative of government incompetence. AI tools turned that narrative into a wildfire, generating variants in multiple languages within hours. Human fact-checkers were left to extinguish flames that kept reigniting in new forms.
The responsibility question remains unresolved. In 2026, European regulators are debating the AI Act and content moderation rules. But the technology evolves faster than legislation. The same AI that can generate a realistic flood simulation can also generate a perfect rebuttal that sounds authoritative. Human judgment is not just bypassed — it is actively targeted.
The Human Role That Remains
Does AI make human judgment entirely superfluous? Not completely. The act of choosing to trust or not trust remains human. The ethical decision to share or withhold still belongs to the person. But these are increasingly reactive choices made within an environment designed by algorithms. The machine sets the agenda.
In the end, the question is not whether AI can produce convincing disinformation. It can. The question is whether society can maintain the human capacity for skepticism when the machine makes every lie look like truth. The heatwave of 2026 will pass, but the AI-generated memories of it will persist, reshaped by algorithms that have no stake in reality.
The boundary between human and machine judgment is not a line — it is a battlefield. And on that battlefield, the human is losing ground not because the machine is smarter, but because it is faster, cheaper, and never gets tired. The cost of being wrong is borne by the people who have to sort truth from endless, automated fiction. That is where AI makes humans superfluous. Not by replacing judgment entirely, but by making careful judgment too expensive for anyone to afford.
Sources
1. Met Office
3. Quota Climat