Whereas it could not include an express political function, expertise is a type of energy. And from the earliest instruments to at this time’s world of social media and generative AI, it comes with main social and political penalties.

Over the course of historical past, too few technologists have actually grappled with this reality. It’s solely lengthy after sprawling chains of unintended penalties ripple out over society (take social media’s function in latest elections, as an example) that they really get up. Now we face a coming wave of transformative expertise led by AI. It is going to have seismic penalties, together with on the very way forward for the very idea of nation-states, however first, we’ll really feel AI’s affect on the subsequent election. We must be one step forward. AI is creating quick–and we’re poorly ready for the affect of this new wave as we hurtle towards a key presidential election.

My concern is that AI will undermine the data house with deepfakes and focused, adaptive misinformation that may emotionally manipulate and persuade even savvy voters and shoppers. What occurs when everybody has the ability to create and broadcast materials with unimaginable ranges of realism? From a politician talking a number of native dialects in India to a collection of doctored movies of Congress members within the U.S., the primary real-life examples are already on the market. And these examples occurred earlier than the flexibility to generate near-perfect deepfakes–whether or not textual content, photos, video, or audio–grew to become as simple as writing a question into Google.

Think about that three days earlier than an election a video of a presidential candidate utilizing a racist slur spreads on social media. The marketing campaign press workplace strenuously denies it, however everybody is aware of what they’ve seen. Outrage seethes across the nation. Polls nosedive. Swing states instantly shift towards the opponent, who, in opposition to all expectations, wins. A brand new administration takes cost. However the video is a deepfake, one so subtle it evades even the perfect fake-detecting neural networks. A grainy, authentic-looking video or an audio recording of a politician defaming a voting bloc could be partaking and convincing. Nonetheless, trusting our eyes and ears is now not doable.

Over the following few years, these applied sciences may have a good wider affect, basically reshaping the stability of energy, shoring up some companies and nations whereas fully undermining others, and rewiring labor markets and safety infrastructures. However forward of those sea modifications is the flood of disinformation round elections. The issue right here lies not a lot with excessive circumstances because it does in delicate, nuanced, and extremely believable situations being exaggerated and distorted. Furthermore, generative AI instruments, for all their undoubted advantages, may very well be weaponized by hostile actors, together with rogue states, introducing new hacking capabilities and systemic vulnerabilities into the guts of the political course of.

Going through these threats, we have to shore up the state and shield society. However initially, we should safeguard the electoral course of, beginning proper now. Free and honest elections are the muse of American society, and subsequent 12 months’s are going to be the primary of the generative AI period. We’re already seeing hints of what AI would possibly do to democracy, intentionally producing pretend info to twist outcomes. That is occurring on American soil and already influencing outcomes. In response, we merely should make sure the integrity of the system is maintained, and this implies explicitly and promptly banning using AI and chatbots in electioneering. These techniques should be stored out of elections, beginning with 2024. No ifs or buts. The democratic course of is just too treasured and too susceptible for a expertise as new and highly effective as AI.

In latest months, because the tide of AI has began to return in, requires regulation have grown from all quarters, together with tech firms themselves. Everybody agrees: Regulating AI is important. However thus far, there hasn’t been enough readability or consensus on the place and find out how to begin. As an alternative, there’s the same old morass of concepts and agendas. Calling for regulation is one factor, entering into the element with specifics is sort of one other. Right here, nonetheless, is a straightforward, clear, and unarguable case for taking fast motion. It’s very important these of us engaged on this expertise state unambiguously what must occur subsequent: banning using AI in elections.

Legislating AI-driven electioneering can be one concrete step in direction of ameliorating the spiraling political penalties of the approaching wave. And it shouldn’t be the final.

Mustafa Suleyman is the co-founder and CEO of Inflection AI. In 2010 he co-founded DeepMind, which was acquired by Google. He’s the creator of The Coming Wave.

The opinions expressed in Fortune.com commentary items are solely the views of their authors and don’t essentially mirror the opinions and beliefs of Fortune.