Andreessen Horowitz’s crypto arm a16z recently published a report on how blockchains could revolutionize the field of artificial intelligence, emphasizing user ownership and open standards. According to a June 11 report, the paper suggested that core context and “agent passports” should be stored in self-custodial wallets, enabling large-language-model interfaces to access preferences instantly without the need for repetitive training.
The report proposed the implementation of a chain-based identity layer that would allow agents to carry verifiable records of owners, capabilities, and payment details across various platforms such as email and Slack. Additionally, systems like World Chain’s World ID could be utilized to authenticate users and screen out bots as generative models continue to proliferate.
Computation and Payment Rails
Regarding computing, the report advocated for decentralized physical infrastructure networks known as DePIN, which would utilize idle GPUs and energy from gaming PCs and spare data center capacity to create permissionless marketplaces. This approach aims to reduce reliance on a small number of cloud providers. On-chain “synchrony layers” would ensure that AI-generated code bases remain aligned, with developers incentivized to patch vulnerabilities through tokenized protocols.
The importance of micropayments was also highlighted, envisioning nano-transactions on low-fee chains that would distribute purchase funds among data sources and AI entities. These rails could also facilitate web crawlers’ access by requiring bots to make token deposits and pay metered fees to website “bouncer” contracts.
Intellectual Property and Companionship
In terms of intellectual property management, the report suggested the use of public on-chain registries to embed ownership metadata directly into creative assets. Smart contract licensing could enable brands to automate charging or permitting derivatives, transforming potential infringement into traceable royalty streams.
Furthermore, advertising could adopt wallet-based zero-knowledge proofs, requiring users to disclose minimal demographic information in exchange for small payments for their attention. Looking ahead, the paper proposed that AI tutors, health aides, and companionship bots should operate under user keys instead of corporate control, leveraging censorship-resistant blockchains and account-abstraction wallets to ensure user privacy and control.
The report concluded by asserting that the integration of blockchains and machine learning could safeguard the open internet by embedding incentives, provenance, and governance directly into the protocol layer, fostering a more secure and transparent digital environment.