The web's broken deal with AI companies 🔗 **Source:** [The web's broken deal with AI companies](https://dri.es/the-webs-broken-deal-with-ai-companies) > [!note]+ Document Notes > tags: [AI, web-economics, content-licensing] > author: Dries Buytaert > source: dri.es > reference: [https://dri.es/the-webs-broken-deal-with-ai-companies] > date created: 2026-01-18 10:30 > > # 🏷️ Title: The web's broken deal with AI companies > <!-- deep-read:sections start --> ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ### 🔗 References - Include links cited in or relevant to the article. ## 📝 Summary - What happened. - AI companies are heavily crawling and training on web content, then answering users directly, undermining the traffic-dependent economics of the open web. - Why it matters. - If creators can’t earn from their work, the supply of high-quality content that AI depends on will erode. - Key stakeholders. - Independent creators, publishers, open source projects, AI companies, intermediaries like Cloudflare, advertisers, and users. - Risks and tradeoffs. - Better UX (instant answers) vs. starving content sources; centralization of power in a few AI/search platforms; over-blocking vs. under-compensating. - Bottom line. - We need enforceable controls on AI crawling plus new content-licensing marketplaces so value is fairly shared between AI firms and web creators. ## 📖 Main Content - Explain the core arguments in plain language. - For decades, the web ran on an informal deal: search engines could index content because they sent users back to creators’ sites, where ads, subscriptions, or leads paid the bills. - Generative AI breaks this deal by training on content and answering inside chat/search interfaces, often without clicks back. This is framed as an extraction problem: AI companies are “takers” siphoning value without reciprocal benefit. - Evidence includes “zero-click” search rising to ~69% with AI Overviews, traffic collapses for sites like Chegg and Stack Overflow, and small creators (e.g., a recipe blogger) losing ad income. - Users love instant AI answers, so the change is driven by genuine UX benefits, not just corporate strategy. But this threatens the sustainability of the open web that feeds AI. - The author argues we must find a new equilibrium balancing great UX with fair compensation for creators. - Existing tools (robots.txt, new CC Signals) are voluntary “social contracts” and not enforceable. - Cloudflare’s new default blocking of AI crawlers is presented as a turning point: a large intermediary enforcing crawler rules and enabling negotiation instead of silent scraping. - Big platforms like Reddit can already strike direct AI licensing deals; small creators cannot. - The author proposes building “content licensing marketplaces” (like Shutterstock + Google Ads) so creators of all sizes can set terms and be paid programmatically for AI training and usage. - The conclusion: AI is not the problem; the lack of economic infrastructure is. Enforceable controls plus marketplaces could make the open web more sustainable than before. - Add sourced context and define terms. - “Zero-click” search: user gets an answer directly on the search results page (or AI answer box) without clicking through to any site. - “Makers vs. Takers”: a framing where some actors invest in producing value (makers) while others primarily extract or arbitrage it (takers) without adequate return. - Cloudflare: a major CDN/security provider sitting in front of ~20% of web traffic, able to inspect and block abusive crawlers. - Content licensing marketplace: a platform where rights-holders set license terms and get paid automatically when their content is used (e.g., Shutterstock for images). - Note uncertainties or contested claims. - Long-term impact of AI-driven traffic reduction: early data suggests harm, but creator and marketer strategies are still adapting. - Whether AI companies can rely indefinitely on frozen or synthetic data is debated; the author argues models need ongoing fresh, high-quality content. - Feasibility and governance of a global content marketplace (pricing, fraud, jurisdiction, privacy) are open design questions. ### 🗒️ Notes - AI is dramatically increasing demand for compute and models; Drupal is investing heavily in AI to keep up with how people will build and consume software. - The open web’s original deal: indexing in exchange for referral traffic, which funded content via ads, subscriptions, or downstream commercial activity. - Generative AI breaks this by: - Training on creators’ content without direct compensation. - Answering inside AI/search interfaces, bypassing creators’ sites. - Data points: - Chegg alleges AI search crushed traffic and revenue (antitrust suit vs. Google). - Stack Overflow’s DAUs and new questions reportedly dropped ~25–50% as devs shift to ChatGPT. - Individual creators (e.g., recipe blogger) see lower traffic and ad income, threatening viability. - Zero-click trend: - After AI Overviews, no-click searches rise from ~56% (May 2024) to ~69% (May 2025). - Fits the author’s earlier “Big Reverse of the Web” thesis: platforms increasingly keep users instead of sending them out. - Some AI-referred users convert better (e.g., >6% conversion vs. 2–3% typical at Acquia), but volumes are very small so far. - New equilibrium: - UX gains from AI are real and desirable. - But extracting value without supporting sources is unsustainable. - Models will need ongoing, evolving content; static scrape sets are not enough. - Current control signals: - robots.txt: widely used but voluntary; non-compliant crawlers can ignore it. - CC Signals: new metadata to indicate permissible AI use; also voluntary/social. - Cloudflare’s move: - Default block for AI crawlers, with opt-in from site owners. - Cloudflare can detect crawlers ignoring robots.txt and block them at scale. - Encodes principle that training data has a price and creators deserve a share. - Large sites like Reddit can already negotiate big direct deals (e.g., $60M/yr with Google); small sites generally can’t. - Missing piece: marketplaces: - Need a scalable system to match millions of creators with AI firms and handle licensing. - Envision: “Shutterstock for training data” + automated, auction-like pricing akin to Google Ads. - Creators set rules/terms; marketplace handles pricing, matching, and payouts. - This levels the playing field so individuals and small orgs can participate. - Structural problem: - AI firms lack scalable ways to find, license, and pay millions of content sources. - Creators lack tools to signal terms, enforce them, and get compensated. - Normative stance: - AI aligns with the web’s mission of democratizing knowledge—but only if it doesn’t economically gut the people who create that knowledge. - There is a “unique opportunity” to co-design economic systems that strengthen the open web instead of hollowing it out. - Author discloses Acquia–Cloudflare commercial ties but frames argument as consistent with prior views on web economics. ## 📚 References - Original article: [The web's broken deal with AI companies](https://dri.es/the-webs-broken-deal-with-ai-companies) - Cloudflare AI crawler announcement (for context): https://blog.cloudflare.com (search “AI crawlers”) - Chegg vs Google antitrust coverage: https://www.reuters.com (search “Chegg antitrust Google AI search”) - Stack Overflow traffic decline & AI: https://stackoverflow.blog or external analyses - CC Signals initiative: https://creativecommons.org/ (search “CC Signals AI”) ## 🗣️ Paraphrase - Paragraph A paraphrase. - AI is rapidly transforming work and creativity, but the way AI companies currently use web content—training on it and answering questions themselves—undermines the open web’s long-standing traffic-for-indexing bargain, so we need mechanisms that fairly compensate creators. - Paragraph B paraphrase. - Voluntary tools like robots.txt and CC Signals are insufficient; infrastructure like Cloudflare’s default AI blocking plus future content licensing marketplaces could enforce creator preferences and enable programmatic, large-scale licensing between AI companies and content owners. ## 📰 Newsletter AI is reshaping not just how we build on the web, but how the web pays its own bills. In “The web’s broken deal with AI companies,” Dries Buytaert argues that generative AI has quietly broken the search-for-traffic bargain that kept blogs, docs, and open source thriving. For the WordPress ecosystem, this has big implications. If AI search answers fewer queries with clicks, ad-based and lead-gen models weaken. But there’s opportunity: infrastructure players like Cloudflare are beginning to default-block AI crawlers, forcing negotiations instead of silent scraping. Dries sketches the next step—a “Shutterstock for training data” powered by programmatic pricing, more like Google Ads than one-off deals. For WordPress creators, hosts, and product teams, now is the time to: - Audit how your content is exposed to AI crawlers. - Decide your stance on training/licensing. - Track early experiments in content licensing marketplaces. Resources: - Article: https://dri.es/the-webs-broken-deal-with-ai-companies - Cloudflare AI crawler policy: https://blog.cloudflare.com - Creative Commons CC Signals: https://creativecommons.org Expect more tools to land in the coming year. The projects that are intentional about their AI/content strategy will weather the “zero-click” era best. ## ✅ Action Items #### DevRel - Developer advocacy - Publish a blog and talk explaining AI crawling, robots.txt, and licensing options for plugin/theme devs. - Demo how to configure Cloudflare (or equivalents) for AI crawler control. - Education - Build a workshop on “AI & the Open Web” covering ethics, licensing, and implementation for WordPress developers. - Schedule office hours on AI policies for docs and code. - Developer experience - Provide sample configs for common hosts (NGINX/Apache, Cloudflare) to manage AI crawlers. - Offer code samples for embedding CC Signals-style metadata in themes/plugins. - Developer marketing - Run a campaign highlighting responsible AI and content protection as a differentiator. - Track conversions from AI-referral traffic separately. - Community management - Start forum threads on AI scraping and content rights; capture FAQs for docs. - Survey maintainers on AI licensing preferences and risk perceptions. - Internal/external/cross-project - Align messaging on AI and licensing across product, legal, and marketing. - Set up feedback loops from community into policy decisions. #### OSPO - Policy & Governance - Draft a policy on whether and how project content may be used for AI training. - Define decision rights for opting in/out of AI licensing deals. - License Compliance - Review compatibility of AI licensing terms with existing open source licenses. - Establish a cadence for training and license scanning focused on AI use. - Security Management - Ensure crawler detection and blocking tooling is secure and auditable. - Document a process for handling abusive or misrepresented AI crawlers. - Software Inventory - Maintain an inventory of repos and docs that are in/out of AI training scope. - Attach licensing and AI-use metadata to SBOMs where applicable. - Developer Support - Provide office hours for questions about AI terms, rights, and licensing. - Create guardrails and checklists for teams negotiating any data deals. - Community Engagement - Explore upstream funding models supported by future content marketplaces. - Consider MoUs with AI providers for fair use and reciprocity. - Strategic Alignment - Define KPIs for sustainable content economics (traffic, funding, licensing revenue). - Report regularly to leadership on AI-related risk and opportunity. #### Community - Maintainers - Host roadmap clinics to assess AI’s impact on project sustainability. - Encourage clear CONTRIBUTING and LICENSE files that mention AI use. - Contributors - Offer “first issue” tasks around adding metadata and signals for AI crawlers. - Pair new contributors with mentors to discuss AI considerations in docs/code. - Sponsors - Communicate how AI-driven traffic losses may affect project budgets. - Share reporting on any licensing revenue or AI partnerships. - Ecosystem - Run cross-project tabletop exercises simulating AI scraping incidents. - Coordinate on responsible disclosure and shared position statements. #### WordPress - Core - Discuss adding standardized mechanisms for AI usage signals in core (e.g., markup, headers). - Consider guidance on default robots.txt patterns for AI crawlers. - Core Performance - Monitor impact of crawler blocking/allowing on performance and infra load. - Track trends in AI-driven traffic vs. traditional search. - Core AI - Define responsible AI guardrails around training on .org content. - Build evaluation sets that respect contributor rights and licensing. - **Core Program Team** - Add AI/content economics as a recurring agenda item in cross-team meetings. - Coordinate experiments (e.g., CC Signals, licensing metadata) across teams. - Mobile - Ensure mobile apps respect AI-related privacy and content policies in features like sharing. - Explore surfacing attribution when content is shared to AI-integrated apps. - Playground - Create labs demonstrating how AI interacts with WordPress sites and logs. - Offer sandbox examples of different AI crawler policy setups. - Community - Add AI/web economics sessions to meetups and WordCamps. - Provide organizers with speaker guides and slide templates. - Training - Develop lesson plans on AI, licensing, and the open web. - Pilot certification modules that include responsible AI practices. - Marketing - Craft narratives around “AI-safe, creator-first web publishing.” - Coordinate social content around major policy or tooling changes. - Support - Add KB articles on AI crawlers, robots.txt, and licensing marketplaces. - Define escalation paths for users hit hard by AI-driven traffic drops. - Design - Explore UX patterns for surfacing licensing/AI-use terms to humans and bots. - Ensure any AI-related banners, notices, or consent flows are clear and accessible. - Documentation - Fill DevHub/HelpHub gaps on AI visibility, robots.txt, and HTTP headers. - Promote examples-first patterns for adding AI signals to themes/plugins. - Photos - Clarify how photos on .org may be used in AI training. - Offer opt-out or specific-license tagging where feasible. - TV - Encourage captions/transcripts with clear licensing for AI usage. - Tag content that may or may not be available for training. - Polyglots - Coordinate translations of AI policy and licensing notices. - Track locale coverage for AI-related docs. - Accessibility - Ensure all AI-related UI and docs meet WCAG 2.2 AA. - Provide guidance for assistive-technology users about AI-driven changes. - Meta (.org) - Implement and maintain AI crawler management across .org properties. - Track and report on AI referral patterns and any licensing experiments. - CLI - Consider commands to manage robots.txt / AI-related headers programmatically. - Add docs and examples for operators. - Hosting - Work with hosts on sensible, pro-creator AI crawler defaults. - Gather telemetry (with consent) on AI traffic vs. classic search. - Tide - Explore whether plugin/theme scores should factor in AI safety or licensing metadata. - Set up feedback loops on false positives around AI-related checks. - Openverse - Clarify AI training terms for media assets. - Implement metadata to signal permitted AI uses. - Themes - Add linting or checklists for including AI-related meta in themes. - Mentor theme authors on communicating rights and licensing clearly. - Plugins - Encourage plugins that help users manage AI crawlers and licensing flags. - Tighten provenance checks when plugins integrate AI features. #### The WPOCC - Programs - Launch fellowships or stipends for maintainers exploring AI-sustainable funding. - Offer microgrants for experiments in licensing metadata and marketplaces. - Projects - Run documentation sprints on AI + open web economics. - Support governance clinics focused on AI data deals. - Funding experiments - Pilot matched grants tied to lost traffic vs. AI licensing revenue. - Use transparent ledgers to track any AI-related funding flows. ### FAIR - Technical independence - Ensure metadata and signals for AI use are locally stored and verifiable. - Encourage signing and provenance tracking for licensed content. - Policies and covenants - Define funding transparency rules for AI-related deals. - Include sunset clauses and anti-capture measures in any agreements. - Productized governance artifacts - Turn AI policies into reusable charters and operator agreements. - Publish scorecards evaluating AI partners on fairness and compliance. ## 🧪 Hypotheses - H1: Implementing enforceable AI crawler blocking (e.g., via Cloudflare) will reduce unauthorized scraping without significantly harming legitimate traffic. - H2: Sites that participate in content licensing marketplaces will see more stable or increased revenue compared to similar sites that rely only on ads/SEO. - H3: Over time, AI models trained solely on older/synthetic data will underperform those that license ongoing fresh web content. - H4: Users will accept slightly more friction (e.g., attribution links, click-outs) in AI answers if they understand it funds creators. - H5: Small and mid-sized creators will adopt standardized licensing tools faster than bespoke contracts when friction is low. ### 📊 Methodology - For H1 - Quasi-experiment: compare traffic and crawler patterns before/after enabling AI blocking across a sample of sites. - Metrics: number of AI-associated user agents, suspected scraping events, organic search traffic. - For H2 - Longitudinal study of sites enrolled in a marketplace vs. a matched control group. - Analyze revenue composition, volatility, and total earnings over time. - For H3 - Benchmark models with different training regimes on up-to-date tasks (news, code, regulations). - Measure accuracy, hallucination rates, and user satisfaction. - For H4 - A/B tests on AI answer interfaces: pure inline answer vs. answer + visible attribution and “read more” clicks. - Track engagement, satisfaction, and click rates. - For H5 - Survey creators and platforms about adoption of standardized metadata/licensing tools. - Track adoption curves as tools become easier and more integrated. ### ❓ Questions - How should content licensing prices be set so they’re fair to both niche bloggers and large publishers? - Who governs a global content licensing marketplace, and how do we prevent capture by a few platforms or AI vendors? - How do open source licenses interact with paid AI training rights? - What attribution norms should exist for AI answers built on licensed content? - How can we give individual creators simple, meaningful control over their AI participation? ### 🔮 Future Research - Economic modeling of various web funding regimes under high AI usage (ads, subs, donations, licensing). - Governance designs for content marketplaces that protect both creators and users. - Technical standards for AI crawler identification, authentication, and reporting. - UX research on how users understand “supporting creators” when using AI tools. - Legal analyses of fair use vs. licensing across jurisdictions. ### 💡 Implications - Practice - Web teams must treat AI crawler management as a core operational concern. - Creators should think beyond ads to licensing and diversified income streams. - Policy - Regulators may need to clarify when AI training is fair use vs. when compensation is owed. - Standards bodies can help define interoperable AI signals and marketplaces. - Security - Stronger crawler enforcement reduces abusive scraping and misrepresentation. - New attack surfaces emerge in licensing marketplaces (fraud, spoofing). - Education - Developers and creators need literacy in AI economics and licensing, not just model usage. - Curricula should cover both ethics and implementation details. - Funding models - AI licensing could become a new revenue pillar for content-heavy projects. - But there’s risk of over-dependence on a few large AI buyers. ### 🧠 Learning Styles Assessment - Reading level estimate - Likely around Grade 10–12 on Flesch–Kincaid: short paragraphs but domain-specific terms (“zero-click”, “training data”, “programmatic pricing”). #### 👀 Visual learners - Present - At least one illustrative hero image; clear section headings. - Improve - Diagrams showing the old search model vs. AI model; flow of value and licensing layers. #### 🔊 Aural learners - Present - No explicit audio, but could be read aloud easily. - Improve - 3–5 minute audio summary or TTS option for the article. #### 🗣️ Verbal learners - Present - Clear definitions and analogies (“Makers and Takers”, “Shutterstock for AI data”). - Improve - Short glossary of AI/web economics terms at the end. #### 🧩 Kinesthetic learners - Present - No direct interactive elements. - Improve - Step-by-step guides for configuring robots.txt/Cloudflare; small experiments with blocking vs. allowing crawlers. #### 🧠 Logical learners - Present - Structured arguments with cause-and-effect (AI Overviews → zero-click → traffic loss). - Improve - Simple tables comparing models (old search vs. AI search vs. marketplace-enabled). #### 🧑‍🤝‍🧑 Social learners - Present - Implicitly invites community and ecosystem responses. - Improve - Discussion prompts or links to community forums about AI and web economics. #### 🧘‍♀️ Solitary learners - Present - Article format supports reflective reading. - Improve - Reflection questions at the end (“How dependent is your site on search traffic?”). #### 🌿 Naturalistic learners - Present - Limited direct environmental analogies. - Improve - Use ecosystem metaphors for web sustainability and resource extraction. ### 📊 Accessibility Assessment #### Screen reader semantics - Present - Semantic headings and straightforward text flow. - Improve - Ensure all images have descriptive alt text; specify language attributes. #### Keyboard navigation - Present - Standard blog layout is usually keyboard-friendly. - Improve - Confirm visible focus states and a skip-to-content link. #### Contrast and readability - Present - Typical blog styling likely meets basic contrast; body text is legible. - Improve - Verify 4.5:1 contrast, comfortable line-height, and font size throughout. #### Text resizing and reflow - Present - Responsive design likely handles zoom up to 200%. - Improve - Test for clipping and horizontal scrolling at high zoom; add prefers-reduced-motion support. #### Media and captions - Present - Static images only; no video/audio to caption. - Improve - If future videos or embeds are added, include captions and transcripts. #### Forms and interactions - Present - Minimal forms (if any) on article page. - Improve - Ensure any comment or signup forms have labels, error messaging, and adequate hit targets. #### Standards reference - Map gaps to WCAG 2.2 AA - Focus on 1.4.x (contrast), 1.4.10 (reflow), 2.1.1–2.1.2 (keyboard), 2.4.x (navigation), 3.3.x (input assistance) as applicable. ### 🌍 DEIB Assessment #### Cultural breadth - Present - Examples from US-centric companies and independent creators. - Improve - Include case studies from different regions and languages. #### Inclusive language - Present - Neutral, professional tone; minimal jargon. - Improve - Add plain-language summaries and define more terms inline for non-technical readers. #### Neurodiversity support - Present - Chunked sections and consistent structure. - Improve - More subheadings, bullet summaries, and explicit key-takeaway boxes. #### Representation - Present - Abstract examples rather than people imagery. - Improve - If adding imagery, ensure diverse, accurately described representation. #### Psychological safety - Present - Calm, non-alarmist framing; acknowledges risks without blame. - Improve - Clear community guidelines and moderated spaces for debate on AI issues. ### ✅ Logical Validity #### Strong reasoning - Inductive/causal reasoning - From observed metrics (zero-click searches rising, traffic declines) to the conclusion that AI changes threaten web economics. Strengthened by multiple examples (Chegg, Stack Overflow, blogger). - Analogical reasoning - “Makers and Takers” and “Shutterstock for AI data” analogies help map known economic patterns to AI. - Normative argument - Asserts that fair compensation is ethically and practically required to sustain content supply; consistent with evidence about dependency on traffic. #### Fallacies - Potential overgeneralization - Limited case studies might not represent all sectors; some niches may benefit from AI. Acknowledge more counterexamples or data segments to strengthen. - Causal complexity - Traffic drops may have multiple causes beyond AI; explicitly noting this would refine the causal claim. #### Tone of voice - Tone - Pragmatic, constructive, cautiously optimistic. - Uses concern for creators and the open web while recognizing user benefits of AI, avoiding alarmism. ### 🛡️ DARVO Analysis - Deny - The article does not deny harms; it highlights them plainly. No DARVO-style denial. - Attack - Critiques AI companies’ behavior as “extraction” but does not personally attack individuals or accusers. - Reverse victim/offender - The piece frames creators as at risk, not AI firms as victims; no reversal of victim/offender roles. - Overall - The rhetoric focuses on systemic incentives and infrastructure rather than blame-shifting; DARVO patterns are not present. ## 🎯 Conclusion The article argues that AI has quietly broken the web’s implicit “traffic for indexing” deal, putting the open web’s economics at risk. Voluntary norms like robots.txt are no longer enough. By combining enforceable crawler controls (e.g., via intermediaries like Cloudflare) with scalable content licensing marketplaces, AI companies and creators could move from extraction to fair exchange. For readers, the next step is to audit how your content is used by AI today and decide what role you want to play in this emerging market. ## 🧱 Refactor Appendix (append after 🎯 Conclusion) ### 📚 Literature Split Pad - Purpose: Stage Konik CEE units that become per-source Literature notes. - Split level: **H3**. - Include verbatim quotes under 25 words and a [view highlight] link. ### [H1] AI licensing can sustain the open web > [!quote] [The web's broken deal with AI companies](https://dri.es/the-webs-broken-deal-with-ai-companies) by [[Dries Buytaert]] via [[dri.es]]. Published 2025-07-01. > “Cloudflare's move is a first step. The next step is building content licensing marketplaces that connect creators directly with AI companies.” Two lines: context and use. Cite data if relevant. [view highlight](https://dri.es/the-webs-broken-deal-with-ai-companies#H1) #xref #articleseed ID:rw_[MISSING] ### [H2] Zero-click AI answers erode web traffic > [!quote] [The web's broken deal with AI companies](https://dri.es/the-webs-broken-deal-with-ai-companies) > “Since Google launched ‘AI Overviews’, the number of searches that result in no click-throughs has increased from 56% to 69%.” Two lines: context and use. [view highlight](https://dri.es/the-webs-broken-deal-with-ai-companies#H2) #xref ID:rw_[MISSING] --- ### 🔖 Atomic Split Pad (Zettel Candidates) - Purpose: Promote reusable claims into Atomic notes. - Split level: **H4**. - Each item: one idea, present tense, specific, testable. #### [Z1] AI licensing marketplaces can fund creators - Under high AI adoption, programmatic licensing provides a new revenue stream for web content. - Links: [[AI-Economics]] [[Web-Sustainability]] #### [Z2] Zero-click AI search threatens site viability - When AI answers reduce outbound clicks, traffic-dependent sites become economically fragile. - Links: [[Search-Ecosystem]] [[Content-Funding]] --- ### ✏️ Fleeting Queue (Extract by Selection) - Purpose: Capture quick actions or sparks for Fleeting notes. - Extraction: **Select one line** → “Refactor: Extract selection to new note (content only).” - Keep each line short and self-contained. - [F1] Map my sites’ exposure to AI crawlers and decide a default stance — https://dri.es/the-webs-broken-deal-with-ai-companies #fleeting #inbox - [F2] Research emerging AI content licensing marketplaces and early adopters — https://dri.es/the-webs-broken-deal-with-ai-companies #fleeting #inbox - [F3] Propose community session on AI, zero-click search, and WordPress sustainability — https://dri.es/the-webs-broken-deal-with-ai-companies #fleeting #inbox --- <!-- deep-read:sections end -->