Opinion by: Casey Ford, PhD, researcher at Nym Technologies

Web3 rolled in on the wave of decentralization. Decentralized applications (DApps) grew by 74% in 2024 and individual wallets by 485%, with total value locked (TVL) in decentralized finance (DeFi) closing at a near-record high of $214 billion. The industry is also, however, heading straight for a state of capture if it does not wake up. 

As Elon Musk has teased of placing the US Treasury on blockchain, however poorly thought out, the tides are turning as crypto is deregulated. But when they do, is Web3 ready to “protect [user] data,” as Musk surrogates pledge? If not, we’re all on the brink of a global data security crisis.

The crisis boils down to a vulnerability at the heart of the digital world: the metadata surveillance of all existing networks, even the decentralized ones of Web3. AI technologies are now at the foundation of surveillance systems and serve as accelerants. Anonymity networks offer a way out of this state of capture. But this must begin with metadata protections across the board.

Metadata is the new frontier of surveillance

Metadata is the overlooked raw material of AI surveillance. Compared to payload data, metadata is lightweight and thus easy to process en masse. Here, AI systems excel best. Aggregated metadata can reveal much more than encrypted contents: patterns of behaviors, networks of contacts, personal desires and, ultimately, predictability. And legally, it is unprotected in the way end-to-end (E2E) encrypted communications are now in some regions. 

While metadata is a part of all digital assets, the metadata that leaks from E2E encrypted traffic exposes us and what we do: IPs, timing signatures, packet sizes, encryption formats and even wallet specifications. All of this is fully legible to adversaries surveilling a network. Blockchain transactions are no exception.

From piles of digital junk can emerge a goldmine of detailed records of everything we do. Metadata is our digital unconscious, and it is up for grabs for whatever machines can harvest it for profit.

The limits of blockchain

Protecting the metadata of transactions was an afterthought of blockchain technology. Crypto does not offer anonymity despite the reactionary association of the industry with illicit trade. It offers pseudonymity, the ability to hold tokens in a wallet with a chosen name. 

Recent: How to tokenize real-world assets on Bitcoin

Harry Halpin and Ania Piotrowska have diagnosed the situation:

“[T]he public nature of Bitcoin’s ledger of transactions […] means anyone can observe the flow of coins. [P]seudonymous addresses do not provide any meaningful level of anonymity, since anyone can harvest the counterparty addresses of any given transaction and reconstruct the chain of transactions.”

As all chain transactions are public, anyone running a full node can have a panoptic view of chain activity. Further, metadata like IP addresses attached to pseudonymous wallets can be used to identify people’s locations and identities if tracking technologies are sophisticated enough. 

This is the core problem of metadata surveillance in blockchain economics: Surveillance systems can effectively de-anonymize our financial traffic by any capable party.

Knowledge is also an insecurity

Knowledge is not just power, as the adage goes. It’s also the basis on which we are exploited and disempowered. There are at least three general metadata risks across Web3.

• Fraud: Financial insecurity and surveillance are intrinsically linked. The most serious hacks, thefts or scams depend on accumulated knowledge about a target: their assets, transaction histories and who they are. DappRadar estimates a $1.3-billion loss due to “hacks and exploits” like phishing attacks in 2024 alone. 

• Leaks: The wallets that permit access to decentralized tokenomics rely on leaky centralized infrastructures. Studies of DApps and wallets have shown the prevalence of IP leaks: “The existing wallet infrastructure is not in favor of users’ privacy. Websites abuse wallets to fingerprint users online, and DApps and wallets leak the user’s wallet address to third parties.” Pseudonymity is pointless if people’s identities and patterns of transactions can be easily revealed through metadata.

• Chain consensus: Chain consensus is a potential point of attack. One example is a recent initiative by Celestia to add an anonymity layer to obscure the metadata of validators against particular attacks seeking to disrupt chain consensus in Celestia’s Data Availability Sampling (DAS) process.

Securing Web3 through anonymity

As Web3 continues to grow, so does the amount of metadata about people’s activities being offered up to newly empowered surveillance systems. 

Beyond VPNs

Virtual private network (VPN) technology is decades old at this point. The lack of advancement is shocking, with most VPNs remaining in the same centralized and proprietary infrastructures. Networks like Tor and Dandelion stepped in as decentralized solutions. Yet they are still vulnerable to surveillance by global adversaries capable of “timing analysis” via the control of entry and exit nodes. Even more advanced tools are needed.

Noise networks

All surveillance looks for patterns in a network full of noise. By further obscuring patterns of communication and de-linking metadata like IPs from metadata generated by traffic, the possible attack vectors can be significantly reduced, and metadata patterns can be scrambled into nonsense.

Anonymizing networks have emerged to anonymize sensitive traffic like communications or crypto transactions via noise: cover traffic, timing obfuscations and data mixing. In the same spirit, other VPNs like Mullvad have introduced programs like DAITA (Defense Against AI-guided Traffic Analysis), which seeks to add “distortion” to its VPN network. 

Scrambling the codes

Whether it’s defending people against the assassinations in tomorrow’s drone wars or securing their onchain transactions, new anonymity networks are needed to scramble the codes of what makes all of us targetable: the metadata our online lives leave in their wake.

The state of capture is already here. Machine learning is feeding off our data. Instead of leaving people’s data there unprotected, Web3 and anonymity systems can make sure that what ends up in the teeth of AI is effectively garbage.

Opinion by: Casey Ford, PhD, researcher at Nym Technologies.

This article is for general information purposes and is not intended to be and should not be taken as legal or investment advice. The views, thoughts, and opinions expressed here are the author’s alone and do not necessarily reflect or represent the views and opinions of Cointelegraph.