# Verifying Reality Onchain *Author: Bankless* *Published: Apr 25, 2025* *Source: https://www.bankless.com/de/verifying-reality-onchain* --- [![](https://storage.ghost.io/c/e4/b7/e4b77544-5a37-4f0b-8824-8440aa348476/content/images/2025/04/unnamed--51-.png)](https://www.bankless.com/sponsor/uniswap-extension-1719068020?ref=read/crypto-infra-truth-trust-ai&email=true) Verifying Reality Onchain Published on Apr. 25, 2025 [ View in Browser ](https://www.bankless.com/mindshare) --- [**Sponsor: Uniswap Labs**](https://www.bankless.com/sponsor/uniswap-extension-1719068020?ref=read/crypto-infra-truth-trust-ai&email=true) — Unichain is live! Bridge & swap with Uniswap Labs' web app or wallet. [Swap Today](https://www.bankless.com/sponsor/uniswap-extension-1719068020?ref=read/crypto-infra-truth-trust-ai&email=true) .  .  . MARKET PULSE Catching Up on DeFAI [![](https://storage.ghost.io/c/e4/b7/e4b77544-5a37-4f0b-8824-8440aa348476/content/images/2025/04/image---2025-04-25T135459.924.png)](https://www.cookie.fun/?type=defai)*[via cookie.fun's DeFAI dashboard](https://www.cookie.fun/?type=defai)* 📸 **Market Snapshot:** Spring seems to have indeed sprung, with the overall AI agents market cap up ~37% over the last 7 days. To sift through all this green, we're going to hone in on a particular area of onchain AI: *DeFAI*. ### **What is DeFAI?** While definitions may differ, I believe the best way to understand DeFAI is as the application of AI in onchain markets, i.e. DeFi, to make using these markets more accessible and efficient. Currently, DeFAI apps mostly take the form of chat-based agent terminals like [**Hey Anon**](https://x.com/HeyAnonai) (+74% this week) or [**Griffain**](https://x.com/griffaindotcom) (+108%) to execute complex, multi-step operations (e.g. swapping then bridging then swapping again) onchain. The ease these platforms provide should not be glossed over, given that current crypto UX makes many people abort when a transaction is required rather than actually sticking around and exploring everything out there. Beyond just increasing ease of use though, DeFAI *also *makes the entire act of using onchain markets more efficient and worthwhile. Take [**Sturdy Finance**](https://x.com/SturdyFinance) (+12%), for instance, an AI-driven yield aggregator whose [average returns far outpace](https://www.sturdysubnet.org/dashboard) the averages of lending hubs like Aave and Maker.  Further, we have a series of projects like [**Taoshi**](https://x.com/taoshiio/status/1911150086064930839) and [**Mode Network**](https://x.com/modenetwork) that seek to use blockchains to surface profitable, financial signals. Certainly a tall order, though, but they are already showing promise—[Taoshi in particular](https://x.com/taoshiio/status/1911150086064930839). Overall, as this intersection evolves, I believe it’ll create a more efficient, accessible, and *profitable* DeFi ecosystem. In the meantime, outside of those mentioned above, make sure to keep tabs on these protocols if you’re interested in DeFAI broadly: - [**Allora**](https://www.allora.network/) — *A self-improving decentralized AI network that lets devs tap collective model outputs for DeFi use-cases.* - [**Almanak**](https://app.almanak.co/home) — *An upcoming network for AI-driven financial agents to optimize and deploy DeFi strategies.* - [**Orbit**](https://x.com/orbitcryptoai) — *A Solana-based agent platform for cross-chain DeFi strategies and automation using $GRIFT.* - [**Wayfinder**](https://www.wayfinder.ai/)  — *An AI “GPS” network that guides agents across chains; uses $PROMPT. * .  .  . NEWS ROUNDUP Crypto as Infrastructure for Truth and Trust in the Era of AI Bankless Author: [ David C](https://www.bankless.com/author/david) [![](https://storage.ghost.io/c/e4/b7/e4b77544-5a37-4f0b-8824-8440aa348476/content/images/2025/04/image---2025-04-25T114245.308-3.png)](https://www.bankless.com/read/crypto-infra-truth-trust-ai)**We're at a strange point in internet history: everyone can publish, but no one can agree on what's real.** AI now generates much of what we see—from news stories and images to entire online profiles. Deepfakes spread fast. Lies outpace corrections. This doesn’t just induce confusion, it comes with real costs: - **$78B** lost annually in 2019 from [disinformation campaigns](https://www.odwyerpr.com/story/public/13448/2019-11-26/cost-fake-news-78-billion.html) - **95%** of elections in 2024 [influenced by fake news](https://upgradedemocracy.de/en/superwahljahr-2024-superjahr-desinformation/) - **70%** of [people distrust the media](https://stylus.com/consumer-attitudes/edelman-trust-barometer-2025-consumers-negative-outlooks) as of 2025 - **26%** of journalists polled in 2022 [unknowingly reported false information](https://www.pewresearch.org/journalism/2022/06/14/journalists-highly-concerned-about-misinformation-future-of-press-freedoms/) The internet once promised access to truth. But at scale, that promise collapsed. Trust evolved from personal relationships to institutional authorities and finally to digital platforms that initially solved the "online trust problem" with reliability and protections.  Yet as these platforms grew into truth gatekeepers, trust became increasingly centralized and fragile. Trusted institutions now feel slow and opaque. The systems we relied on haven't kept up with today's information flows. This is where crypto starts to matter in a new way. While initially designed to create trustless transactions, crypto has evolved into belief-coordination infrastructure—enabling strangers to reach consensus through transparency, incentives, and decentralization. These values are being repurposed—from finance to information—to answer two core questions of the AI age: *What's real?* And *who should I trust?* ## **Truth: How Blockchains Help Us Know What's Real** The web is saturated with content but hollowed out in confidence. Deepfakes, hallucinated text, and manipulated screenshots blur the line between fact and fiction. The issue isn't just misinformation. Instead, the mechanisms for verifying reality are outdated and untrusted. Here, cue in **blockchains**, which provide ways to tackle the erosion of truth at scale, using public ledgers and **open markets** to chronicle and surface what is real or provably happened. They provide us avenues to know something is real, using decentralized discovery processes to generate reliable signals, not through institutional authority, but through participation, transparency, and economic incentives. One of the clearest examples is provenance. Blockchains timestamp and preserve the origin of information. This is why **NFTs** prove compelling: they make digital media *traceable*. A file can be cryptographically linked to its creator, restoring value to originality in an internet designed for copying. Now that model is expanding. Consider projects like: - IP blockchain [**Story Protocol**](https://www.bankless.com/read/what-story-protocol-built) looks to complement copyright law with a decentralized system that lets creators anchor their IP onchain with embedded attestations, proving originality, authorship, and compliance. - Move-based L1 [**Sui**](https://www.bankless.com/read/whats-next-for-sui-in-2025) has a design that's relevant here, too. Unlike typical chains that simply track ownership, Sui’s object-based system treats each digital item as an object carrying its own permanent history and record. Coupled with decentralized storage service [Walrus](https://x.com/WalrusProtocol) and the security protocol [SEAL](https://x.com/Mysten_Labs/status/1908206035493343606), this full-stack design makes authenticity *native *to the digital experience rather than something that must be taken on faith. Then, while blockchains’ public ledgers fix *origin *for maintaining truth, their open markets fix *incentives *to surface it: - [**BitMind**](https://x.com/BitMindAI), a subnet on the decentralized AI platform [**Bittensor**](https://www.bankless.com/profile/bittensor), uses crypto incentives to drive a system for identifying deepfakes. Miners and validators compete to analyze [media for authenticity](http://a/), earning tokens based on correct detections. - **Prediction markets **offer another powerful approach to surfacing truth. Beyond just headline-worthy success during the election, [**Polymarket**](https://polymarket.com/)'s predictions achieve ~90% accuracy a month before events, and 94% four hours before, according to [data scientist Alex McCullough](https://dune.com/alexmccullough/how-accurate-is-polymarket). While prediction markets don't guarantee truth, they impose powerful economic incentives that sharpen reasoning, making profits for accuracy and threatening costs for errors, producing strong reality approximations when liquidity is deep, though they remain vulnerable to manipulation in thinner markets. Still, they scale scrutiny better than legacy systems. [![](https://lh7-rt.googleusercontent.com/docsz/AD_4nXdHgyoKmT5WbrY2lUdJcxHj6Qxe7ezfymqPaBWQJsgAAQd9XVkMsWZEeab1J2Jf6gVnQZSQSS5rm2AXajGiUpuLm3Ve5GETW5FXD6Uf2tyO8VGVyA8G5yyP_FAd1h5iuQmqIXWw9Q?key=ZVpsKS-qb1shVqsDvwGKoWPe)](https://dune.com/alexmccullough/how-accurate-is-polymarket) ## **Trust: How Blockchains Help Us Know Who to Trust** But knowing what's real is only part of the challenge. Just as important is deciding who to rely on when facts aren't obvious, actors are pseudonymous, or signals are mixed. In this vein, blockchains can help establish trust using public records, shared credentials, and open, auditable systems. This starts with infrastructure. Decentralized identity systems take multiple approaches to solving the "who's real" problem. [**Worldcoin**](https://x.com/worldcoin) uses iris scanning for biometric verification, [**Humanity Protocol**](https://www.humanity.org/) leverages palm prints as secure identity keys, while [**Billions Network**](https://x.com/billions_ntwk) offers a simpler solution using just passports and phones without specialized hardware. Together with technologies like decentralized identifiers, verifiable credentials, and soulbound tokens, these systems ask the right question for the AI era: *how do we know there's a person on the other side?* That question is only becoming more urgent. AI isn't just smart; it's persuasive. As OpenAI's Sam Altman warned, AI may become superhuman at persuasion before it becomes superhuman at intelligence. At the social level, crypto's systems can better simulate the nuance needed for establishing trust. [**Intuition**](https://www.intuition.systems/), for example, is building a kind of "Web of Trust"—where users endorse each other and build layered reputation graphs. This structure captures how trust is personal and contextual. A developer might have no resume, but deep DAO respect. A person might lack a credit score, but be a pillar of their community. Crypto systems give that trust form and mobility. [![](https://storage.ghost.io/c/e4/b7/e4b77544-5a37-4f0b-8824-8440aa348476/content/images/2025/04/image---2025-04-25T110838.071.png)](https://x.com/0xIntuition/status/1910087888894869605)These systems are promising, but they're not magic. Prediction markets can be gamed. Attestations can be noisy. Identity protocols are still emerging. But here's the key difference: crypto systems can evolve. If a centralized platform breaks, users are stuck. If a decentralized one breaks, the community can fork it, fix it, and improve it—all out in the open. As AI reshapes the content landscape, crypto protocols are starting to fill the vacuum: - **Truth** is no longer a fixed announcement — *it's a public process* - **Trust** is no longer issued from the top — *it's earned, contextual, and portable* These are infrastructural shifts, not just technical ones. They change how we verify, who we believe, and what we can build together. If this direction holds, crypto won't just run apps—it will help structure belief in our AI age. [Bookmark on Bankless](https://www.bankless.com/read/crypto-infra-truth-trust-ai) --- ## ***Plus, other news this week*...** ### 🤖 **AI Crypto ** - **Billy Bets** [announced its 2.0 betting framework](https://x.com/BillyBets_ai/status/1915104808513474733) - **Eliza Labs** [opened its auto.fun launchpad](https://x.com/autodotfun/status/1913705865770619285) - **Mirra **[launched its data curation platform](https://x.com/MirraTerminal/status/1915059420548956306) - **Paradigm **[led Nous Research's $50M Series A](https://www.theblock.co/post/352000/paradigm-leads-50-million-usd-round-decentralized-ai-project-nous-research) - **Smolverse **[released agent gamer AI Smols](https://x.com/smolverse/status/1914514380479455533) - **THINK** [dropped its Think Agent Bundle NFTs on Magic Eden](https://x.com/MagicEden/status/1914711299054768452) - **Virtuals** [introduced its Yap for Points system](https://x.com/virtuals_io/status/1915051042095726853) - **Zerion **[unveiled Zerion Copilot](https://x.com/zerion/status/1914325864239751214) ### 📣 **General News** - **Adobe **[released new Firefly image gen models](https://techcrunch.com/2025/04/24/adobe-releases-new-firefly-image-generation-models-and-a-redesigned-firefly-web-app/) - **Character.AI** [unveiled its AvatarFX video model](https://techcrunch.com/2025/04/22/character-ai-unveils-avatarfx-an-ai-video-model-to-create-lifelike-chatbots/) - **OpenAI** [expanded access to its deep research feature](https://x.com/OpenAI/status/1915505959931437178) - **xAI **[announced the launch of Grok Vision](https://techcrunch.com/2025/04/22/xais-grok-chatbot-can-now-see-the-world-around-it/) ### 💻 **Vibe Coding** - **0xDesigner **[published a guide on building with Cursor](https://x.com/0xDesigner/status/1915152588913422778) - **Lens **[hosted a vibe coding livestream talk with the Avail team](https://x.com/LC/status/1915435757222646260) - **Farcaster** [released a quickstart primer on building Mini Apps with AI](https://farcaster.blog/build-farcaster-mini-apps-with-ai) - **Greg Isenberg** [outlines a GPT4.1 vibe code prompting guide](https://x.com/gregisenberg/status/1912298263987757296) - **Greg** [also shares a "practical" guide for vibe coding with Cursor](https://x.com/gregisenberg/status/1914302781168537702) ### 📚 **Reads** - [Masa: AI's Data Tap](https://x.com/0xPrismatic/status/1914844415128166433) — Teng Yan - [So I Clanked a Project](https://paragraph.com/@matt/so-i-clanked-a-project) — Matt - [Latent Reflections 001](https://paragraph.com/@eclecticcapital.eth/latent-reflections-001) — Natalie - [The Next Crypto AI Meta](https://x.com/Defi0xJeff/status/1914299816315937005) — 0xJeff ---