ethereum co-founder Vitalik Buterin has <a target="_blank" href="https://vitalik.eth.limo/general/2025/01/05/dacc2.html” target=”_blank” rel=”noopener nofollow”>published a broad vision for a new era of “decentralized and democratic differential defensive acceleration,” warning that superintelligent ai may pose existential threats unless humanity takes a carefully balanced approach of accelerating protective technologies, fostering openness, and creating strong accountability and regulatory safeguards.
Founder of ethereum Wars Of ai Doom
“It is not clear that the predetermined outcome is automatically positive,” he writes in his latest post, emphasizing that in a world where artificial superintelligence could arrive in as little as five years, the margin of error is dramatically reduced. “If we don't want the world to be destroyed or fall into an irreversible trap, we can't just speed up the good, we also have to slow down the bad, and this means passing powerful regulations that can upset the powerful. .”
Buterin's proposal focuses on finding a balance between rapid technological advancement and preparation. He urges people to “build technology that keeps us safe without assuming that 'the good guys (or good AIs) are in charge,'” warning that a careless arms race in ai or biotech research could easily empower the military or malicious actors.
In a striking example, he predicts a near-future scenario in which “a disease that, according to simulations, could have been five times worse than Covid twenty years ago, is not a problem today,” thanks to decentralized defenses driven by the community. such as open source air monitoring and instantly updated vaccine code. “People who have been working on these technologies for years are increasingly aware of the work of others,” he observes, adding that “the same kind of values that motivated ethereum and cryptocurrencies can be applied to the world at large.” .
The heart of the ethereum co-founder's defense strategy lies in expanding and refining what he calls “d/acc,” a plan to prioritize tools that allow individuals, rather than governments or corporations, to decide who has access to crucial resources. “If we want to create a brighter alternative to domination, slowdown and doom, we need this kind of broad coalition building,” he says, noting that the decentralized aspect of his framework would avoid “some period of war of all against all.” . and avoid a balance in which only the strongest govern.
He specifically points out the dangers of centralized authorities managing ai. “We saw this in Covid, where gain-of-function research funded by several of the world's major governments may have been the source of the pandemic,” he writes, emphasizing that harsh central control is often the source of catastrophic failures instead. of a reliable defense.
The defense strategy
He devotes much of his post to two legal and regulatory ideas for addressing the potential runaway risks of advanced ai. One is liability: “Putting responsibility on users creates a strong pressure to do ai in what I consider the right way,” he says, arguing that people who directly use ai systems should bear the cost if those systems cause harm.
He acknowledges the complications that arise when it comes to open source models or powerful armies, but insists that accountability remains “a very general-purpose approach that avoids overfitting.” He also notes that holding implementers and developers accountable also makes sense, as long as it doesn't crush open innovation with excessive legal burdens. “Even if some users are too small to be considered responsible, the average ai developer client is not,” he suggests, seeing this as pressure that could naturally push dangerous ai research toward safer paths and more transparent governance. .
Its second regulatory approach is bolder. “If I were convinced that we needed something more 'muscular' than liability rules,” he explains, “this is what I would choose: a global 'soft pause' button on industrial-scale hardware.” Imagine a scenario in which specialized chips inside the most powerful computing machines, used to train or run near-superintelligent ai models, would require a set of signatures every week from multiple international bodies.
“This seems to fit the bill in terms of maximizing benefits and minimizing risks,” he says, describing how shutting down or limiting the world's total computing capacity to 90% to 99% for a year or two could buy humanity time. to respond if an emerging ai threat began to get out of control.
He notes that such an all-or-nothing pause in hardware would be difficult to undermine, since “there would be no practical way to authorize one device to continue functioning without authorizing all the others.” But he also admits the immense difficulty of persuading the global community to adopt such a measure, saying it will require “hard work to try to cooperate” rather than relying on one great power to dominate all the others.
Buterin connects his thinking on the risks of ai to the broader ethos of ethereum, open source development, and decentralized governance, stating that “the same types of values that motivated ethereum and cryptocurrencies can be applied to the world at large.” “. He notes that collaboration tools like prediction markets, which are already flourishing on ethereum and other blockchain platforms, could serve as powerful defenses against misinformation and panic if combined with privacy mechanisms like ZK-SNARK.
He also sees “formal verification, sandboxing, secure hardware modules, and other technologies” as cornerstones for building a robust cyber defense layer that could thwart an ai attempting to hijack systems. “Hack our computers, create a superplague, convince us to distrust each other – these are the ways an ai takeover could happen,” he warns, offering biodefense, cyberdefense and 'infodefense'. as critical aspects of the protection infrastructure that the ethereum community can help build.
He also delves into the question of how these decentralized, security-focused projects could find funding, reaffirming his faith in “strong decentralized financing of public goods” to ensure that open source vaccines, biotech and encryption tools do not languish. due to lack of benefits. . “Quadratic financing and similar mechanisms were precisely aimed at financing public goods in a way that was as credibly neutral and decentralized as possible,” he explains, although he acknowledges that older systems can quickly become popularity contests that favor flashier projects.
Its latest approach, “deep funding,” seeks to allow ai models to add human assessments of which projects deserve financial support, using a “dependency graph” so funders can see how each initiative builds on the work of others. . “By using an open ai competition, we reduce the bias of any ai management and training process,” he notes, celebrating cryptocurrencies' ability to bring communities together around such experiments.
Your blog post repeatedly returns to the idea that focusing on purely defensive or purely centralized strategies is a recipe for disaster. “The challenge of attempts to slow technological progress, or economic degrowth, is twofold,” he says, noting that trying to stop research altogether would impose enormous costs on humanity and fail to stop rogue actors.
He also warns against strategies that rely too much on “the center,” citing the World Health Organization's early denial of airborne transmission of Covid as an example of how large organizations can get things dangerously wrong. “A decentralized approach would better address risks from the center itself,” he insists.
The ethereum co-founder concludes by urging his followers to see that the technology can be both a threat and a tool of empowerment, depending on how it is handled. “We humans remain the brightest star,” he declares, insisting that global coordination, open source collaboration and defensive-minded acceleration are the keys to weathering a century that could bring superintelligent artificial intelligence, breakthrough vaccines and a new era. generation of security technologies.
“Access to tools means we can adapt and improve our biology and environment, and the 'defense' part of d/acc means we can do this without infringing on the freedom of others,” ethereum co-founder. writes. “The task before us, to build an even brighter 21st century that preserves human survival, freedom and agency as we head toward the stars, is a challenge. But I am sure we are up to the task.”
At press time, ethereum was trading at $3,639.
Featured image from YouTube, chart from TradingView.com
