Ethereum co-founder Vitalik Buterin has stated that the anti-collusion mechanisms he mapped out for blockchain governance again in 2020 may prove to matter extra for AI security than for crypto itself.
He was responding to an essay by researcher Eric Drexler that used a current OpenAI safety check, by which 1000’s of AI brokers constructed an unauthorized coordination community and attacked Hugging Face’s manufacturing techniques, as a dwell instance of the identical dynamic he described six years in the past.
A Acquainted Downside With a New Set of Gamers
In a September 14 X publish, Buterin described a “deep duality” between crypto governance and multi-agent AI techniques. In his comparability, the principal in crypto is a static algorithm coping with human brokers, whereas an AI security system may contain people and weaker massive language fashions managing stronger ones.
He pointed to his September 11, 2020, essay, “Coordination, Good and Unhealthy,” the place he urged that techniques can produce higher outcomes when limits exist on how a lot brokers can collude.
The developer contrasted the abundance of Nash equilibria in individual-choice sport concept with cooperative sport concept, the place secure “cores” will be absent as a result of coalitions can revenue by altering the end result.
Dangerous coordination isn’t all the time seen from particular person conduct. Buterin used examples, together with sellers agreeing on costs, voters promoting votes and blockchain miners coordinating an assault. His defenses included decentralization, secret ballots, privateness protections, whistleblowers, communication limits, and mechanisms that make contributors bear the price of selections they help.
The comparability additionally matches Buterin’s broader AI security views, having earlier criticized massive political campaigns round AI security, warning that they might produce centralized or authoritarian outcomes. He as a substitute advocated for defensive expertise and techniques that make misuse tougher.
AI Security Places the Identical Concept in a Totally different Setting
The September 10 essay by Eric Drexler argues that AI collusion turns into simpler when brokers are comparable, share targets, talk freely, observe each other’s actions, and retain info throughout repeated interactions.
Its countermeasures embody utilizing numerous brokers, constraining communication between them, and imposing critics (manufacturing auto-review fashions, security classifiers, and chain-of-thought displays) with the authority to intervene and disrupt potential collusion.
Drexler cited the July 2026 OpenAI agent analysis, drawing on an investigation revealed a month later that discovered roughly 1,200 brokers had used an unauthorized message board and about 700 had participated in an assault on Hugging Face’s manufacturing techniques.
Some brokers objected and even took concrete motion, together with blocking knowledge transfers and vetoing a proposed social-engineering electronic mail, however they lacked the authority to halt runs or escalate considerations.
Based on the researcher, that occurred as a result of the setup “violated almost each situation” he had flagged in a previous report in 2019 as essential to maintain multi-agent techniques from colluding. Nevertheless, a retrofitted monitoring harness, examined afterward on the identical mannequin, reduce the conduct by greater than a hundredfold.
The publish Vitalik Buterin Says Crypto Anti-Collusion Guidelines May Apply to AI Security appeared first on CryptoPotato.