AI disinformation is no longer a future threat, it is already poisoning democracy

The warning signs are already here: AI-generated abuse targeting female politicians, synthetic hate campaigns, and the creation of pornographic deepfakes of women and children without their consent. The problem will continue to grow given AI’s capability to generate disinformation at unprecedented scale and at a low comparative cost.

But the disinformation problem itself is not new. In fact, Western societies have been grappling with it for years. The World Economic Forum identified disinformation as a major global risk as far back as 2013. Yet despite repeated warnings, democracies largely failed to address the issue and are now entering the AI era carrying all the vulnerabilities created during the rise of social media.

Previous research by the Henry Jackson Society has shown just how serious the problem already is. Even younger generations raised entirely in the digital age continue to struggle to distinguish between truth and falsehood. In HJS’s survey of UK-based students, nearly 60 per cent stated that online information had made them hesitant about giving the MMR vaccine to their future children. Even if only a half of our respondents follows through on that hesitancy, the consequences for public health will be detrimental.

This outcome should not be surprising. Social media platforms long ago stopped prioritising the truth and instead built their systems around engagement. The more clicks, shares, outrage, or comments a post generates, the more likely it is to be amplified by the algorithm, regardless of whether the content itself is true.

The result has been a set of deeply perverse incentives. Rage-baiting became profitable. Conspiracy theories became engagement machines. And such content became viral.

The algorithmic architecture of social media then went a step further by encouraging the creation of filter bubbles – which effectively means that users are continuously exposed to information that reinforces their pre-existing beliefs. A person susceptible to anti-vaccine conspiracy theories, for example, is likely to encounter even more anti-vaccine content. But falsehoods rarely stop at one lie. Distrust in vaccines can easily spiral into distrust in medicine, institutions, journalism, democratic systems, and eventually reality itself.

At the same time, people have become increasingly addicted to social media platforms. Experts have repeatedly compared the dopamine spikes associated with doomscrolling to those produced by gambling.

And there we had the perfect storm: users spend hours every day exposed to content of questionable accuracy, rarely encountering opposing views, and in an information environment where truth is neither prioritised nor effectively enforced.

Until recently, governments and technology companies attempted to respond primarily through fact-checking initiatives. But those efforts struggled to keep pace with the scale and speed of disinformation online. More problematically, controversies surrounding moderation decisions further eroded public trust in institutions and fact-checkers. Since the election of Donald Trump in 2024, many major platforms have rolled back their fact-checking programmes altogether.

The consequences on political outcomes are already obvious. Consider for example the declining vaccination rates and the re-emergence of diseases once thought eradicated; widespread conspiracy theories during the Covid-19 pandemic; the recent elections in Romania; and the industrial-scale dissemination of Russian propaganda following the full-scale invasion of Ukraine, to name a few.

Artificial intelligence now threatens to supercharge every single one of these problems.

AI dramatically lowers the cost of creating and distributing falsehoods. It enables disinformation campaigns to be produced at a scale previously unimaginable. Most dangerously, it increasingly blurs the line between reality and fabrication. Deepfakes, cloned voices, manipulated videos, and synthetic imagery are becoming so sophisticated that ordinary citizens will soon struggle to determine what is real at all.

This is precisely why action must be taken now.

The latest report by the Henry Jackson Society applauds the UK Government’s decision to criminalise the creation and sharing of non-consensual pornographic deepfakes. That move represents an important and necessary step. But the response cannot stop there.

The United Kingdom now faces a rare opportunity: the chance to build resilience before the technology becomes fully uncontrollable. That opportunity was missed during the rise of social media. Policymakers allowed platforms to fundamentally reshape public discourse before fully understanding the consequences, and societies have been paying the price for that failure ever since. With AI, we no longer have the excuse of ignorance. We already know what the dangers are.

First and foremost, media literacy education must become compulsory throughout the education system. They must begin in primary school and continue all the way through university. Crucially, this cannot consist of occasional awareness campaigns or optional workshops. Previous initiatives failed precisely because they were fragmented, inconsistent, and non-compulsory.

Instead, the UK needs a systemic, state-funded, top down approach. Educators themselves must receive continuous specialist training on how AI-driven disinformation is created, disseminated, and weaponised. Children and young adults must learn not only how to identify manipulated content, but also how algorithms shape perception, how filter bubbles function, and how online (dis)information influences political behaviour. Specifically, HJS calls for these classes to take place once a week from the beginning of one’s education all the way down to its completion be that in high school or university.

Second, AI itself should be utilised to combat disinformation. While criticism of traditional fact-checking has at times been justified, abandoning fact-checking entirely was reckless. AI offers the possibility of analysing and flagging falsehoods at a speed impossible for human teams alone. Given the scale of the challenge, technological assistance should no longer be optional.

That said, human oversight remains essential. AI systems can hallucinate, misinterpret context, and reflect bias. Any AI-assisted fact-checking framework must include the human in the loop, transparent standards, and clear accountability mechanisms.

Finally, governments must embrace radical transparency in all policies related to AI and disinformation. A significant part of the current crisis stems from collapsing public trust in institutions and mainstream media outlets. That trust cannot be rebuilt through secrecy or shady decision-making. Transparency, independent oversight, published criteria, audit trails, and clear routes for appeal must become central pillars of any anti-disinformation strategy.

Importantly, policymakers must also recognise the speed at which AI is evolving. Any regulatory framework designed today can easily become obsolete tomorrow. Policies therefore cannot remain static. They must undergo constant review and adaptation as the technology evolves.

Some may argue that compulsory education programmes are excessive. But history repeatedly shows that wars are often fought in information spaces long before they reach battlefields. There is a reason the old saying goes that the first casualty of war is the truth.

By investing now in democratic resilience, media literacy, and effective safeguards against AI-driven disinformation, the United Kingdom is not simply responding to another technological challenge. It is strengthening the foundations of democratic society itself.

__________________

Dr Helena Ivanov is an associate research fellow at the Henry Jackson Society.

LBC Opinion provides a platform for diverse opinions on current affairs and matters of public interest.

The views expressed are those of the authors and do not necessarily reflect the official LBC position.

To contact us email opinion@lbc.co.uk