Democracy and the open internet die in daylight
🔗 **Source:** [Democracy and the open internet die in daylight](https://heatherburns.tech/2025/10/22/democracy-and-the-open-internet-die-in-daylight/)
> [!note]+ Document Notes
> tags: [Journalism, Adtech, Agentic AI]
> author: Heather Burns
> source: heatherburns.tech
> reference: [https://heatherburns.tech/2025/10/22/democracy-and-the-open-internet-die-in-daylight/]
> date created: 2025-10-22 00:00
>
> # 🏷️ Title: Democracy and the open internet die in daylight
>
<!-- 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: The Washington Post offered “free access” to stories if readers download Perplexity’s proprietary Comet browser, which is built around agentic AI.
- Why it matters: Ties access to news to a closed, privacy-risky browser and signals a shift from ads to AI-driven funding.
- Key stakeholders: News publishers, Perplexity/AI platforms, readers, privacy and open web advocates.
- Risks and tradeoffs: Browser lock-in, privacy erosion, centralization of distribution, displacement of adtech with another extractive model.
- Bottom line: Replacing adtech with agentic AI—especially via proprietary gatekeeping—threatens the open web and user rights.
## 📖 Main Content
- Core argument: Conditioning news access on a proprietary, AI-centric browser is a “1990s throwback” that swaps one harmful funding model (adtech) for another (agentic AI) without addressing privacy and rights.
- Context/definitions:
- Agentic AI: Autonomous assistants that browse/summarize/act for users, often requiring extensive data collection.
- Funding shift: From surveillance ads to AI referral/aggregation deals and vertically integrated distribution (AI browser).
- Uncertainties:
- Exact business terms between publishers and AI vendors.
- Long-term viability of AI-driven payoffs versus subscriptions or public-interest funding.
### 🗒️ Notes
- “You get free access to news if you download PerplexityAI’s proprietary browser, which is built entirely around agentic AI.”
- “Day 0 of an experiment in shifting the funding model of journalism from adtech to agentic AI.”
- “It sure as fuck isn’t trading one privacy-obliterating technology for another, packaged as opportunity.”
- Author frames self as a UK tech policy advocate for “human rights, privacy, accessibility, and freedom of expression.”
## 📚 References
- Perplexity’s Comet announcement (context for agentic browser): https://www.perplexity.ai/
- Background on adtech harms (privacy/security): https://www.eff.org/issues/online-tracking
- Open web principles (standards/human rights): https://www.w3.org/Consortium/mission
## 🗣️ Paraphrase
- Paragraph A: A major newspaper dangled free access in exchange for installing a proprietary AI browser—an alarming echo of old walled-garden tactics.
- Paragraph B: Journalism’s adtech dependence helped break the ecosystem; replacing it with agentic AI doesn’t fix privacy or power asymmetries.
## 📰 Newsletter
WordPress folks: a notable publisher just tested “free access if you install an AI browser.” This is the new lock-in: swapping adtech for agentic AI distribution. For site owners:
- Audit privacy policies and consent flows.
- Evaluate server-side analytics and subscriptions over surveillance ads.
- Track AI agents in logs; deploy robots and AI-specific access policies thoughtfully.
Resources: WP consent tools (Complianz), privacy-first analytics (Plausible), AI crawler guidance (Search Console + robots best practices).
## ✅ Action Items
#### DevRel
- Launch demos on privacy-first analytics and subscription/paywall integrations.
- Publish SDK examples for transparent AI crawler handling and attribution.
- Run workshops on “AI agents vs. robots.txt vs. paywalls” with measurable adoption.
#### OSPO
- Draft AI agent access policy and contribution guidelines for open web standards.
- Establish license/terms reviews for AI integrations; training on data provenance.
- Define SBOM for tracking AI-related dependencies.
#### Community
- Host governance clinics on funding models beyond adtech/AI lock-in.
- Pair newcomers on privacy issues; publish moderator guidelines for AI content.
#### WordPress
- Core: explore hooks for agent/crawler identification and rate-limiting.
- Core Performance: measure AI-agent traffic cost; budget accordingly.
- Core AI: guardrails for privacy, eval sets for hallucination risk.
- Core Program Team: coordinate with Meta and Docs on AI traffic guidance.
- Playground: labs demonstrating robots, paywall headers, and consent UX.
- Training/Docs: patterns for subscriptions, memberships, reader revenue.
- Accessibility: ensure consent/paywalls meet WCAG; avoid dark patterns.
- Meta/Hosting: cross-host defaults for bot filtering and log visibility.
- Plugins/Themes: add provenance checks, signature/attestation for AI-injected content.
#### The WPOCC
- Fund maintainers for privacy/security features; publish impact reports.
- Pilot matched grants for open revenue experiments (subs/donations).
### FAIR
- Technical independence: provenance headers, signed feeds, agent access logs.
- Policies: funding transparency, anti-capture covenants, sunset clauses on deals.
- Governance artifacts: standard operator agreements and scorecards for AI access.
## 🧪 Hypotheses
- Tying access to a proprietary AI browser reduces user privacy relative to ad-supported access.
- Agentic AI distribution increases publisher dependency on a few vendors.
- Reader trust declines when access is conditioned on closed tools.
- Privacy-first revenue (subs/donations) can match or exceed AI-referral deals over 12–24 months for mid-sized publishers.
### 📊 Methodology
- Compare telemetry: cookie syncs, device fingerprinting, data sharing under each model.
- Market structure analysis: HHI for AI distribution channels.
- A/B trust surveys with disclosed access conditions.
- Revenue cohort study: subscription/donation funnels vs. AI referral revenue.
### ❓ Questions
- What data does the AI browser collect and share for “free access”?
- How are attribution and payments calculated between AI vendor and publisher?
- What exit ramps exist if the AI vendor changes terms?
### 🔮 Future Research
- Standards for AI agent identification and rate control.
- Privacy-preserving paywall and metering protocols.
- Public-interest funding models scalable beyond top-tier outlets.
### 💡 Implications
- Practice: prioritize open standards, subscriptions, and privacy-first analytics.
- Policy: transparency mandates for AI distribution deals and data flows.
- Security: monitor automated agent traffic and abuse vectors.
- Education: reader literacy on tradeoffs of “free” via proprietary tools.
- Funding: diversify revenue away from ads and AI lock-in.
### 🧠 Learning Styles Assessment
- Reading level: ~Grade 10–11 (Flesch-Kincaid estimate).
#### 👀 Visual learners
- Present: screenshots described.
- Improve: flow diagram of adtech → agentic AI pipeline.
#### 🔊 Aural learners
- Improve: 3-minute audio summary.
#### 🗣️ Verbal learners
- Present: clear stance and definitions.
- Improve: glossary (agentic AI, lock-in, metering).
#### 🧩 Kinesthetic learners
- Improve: hands-on checklist to audit AI agent traffic.
#### 🧠 Logical learners
- Improve: decision tree for funding model choices.
#### 🧑🤝🧑 Social learners
- Improve: moderated forum prompts on alternatives to adtech.
#### 🧘♀️ Solitary learners
- Improve: reflection prompts on privacy tradeoffs.
#### 🌿 Naturalistic learners
- Improve: sustainability lens on compute and bandwidth costs.
### 📊 Accessibility Assessment
#### Screen reader semantics
- Improve: ensure alt text for screenshots; language metadata.
#### Keyboard navigation
- Improve: verify focus order and skip links on related pages.
#### Contrast and readability
- Present: likely acceptable; confirm 4.5:1 body text.
#### Text resizing and reflow
- Improve: test at 200% zoom; prefers-reduced-motion.
#### Media and captions
- Improve: add audio descriptions and transcripts for any embeds.
#### Forms and interactions
- Improve: explicit labels, inline validation for consent/paywall UI.
#### Standards reference
- Map fixes to WCAG 2.2 AA: 1.4.*, 2.4.*, 3.3.*, 2.5.*.
### 🌍 DEIB Assessment
#### Cultural breadth
- Improve: examples beyond US/UK publishers.
#### Inclusive language
- Present: clear, direct; define jargon.
#### Neurodiversity support
- Improve: chunked sections and consistent callouts.
#### Representation
- Improve: cite diverse voices on open web governance.
#### Psychological safety
- Present: strong critique; remain respectful; add community guidelines for discussion.
### ✅ Logical Validity
#### Strong reasoning
- Inductive causal: proprietary AI browser conditions access → privacy and lock-in risks; aligns with known patterns of walled gardens.
#### Fallacies
- Potential hasty generalization: one instance extrapolated to industry shift; mitigate by adding trend data across outlets.
#### Tone of voice
- Forthright, critical, advocacy-oriented (e.g., “grinds teeth into powder”).
### 🛡️ DARVO Analysis
- Deny: Not applicable; no denial pattern by the author.
- Attack: Critique targets a model/decision, not individuals.
- Reverse victim/offender: Warns that users/public become victims of lock-in; not framed as victimhood by the critic.
## 🎯 Conclusion
- Bottom line: Swapping adtech for agentic AI via proprietary browsers endangers privacy and the open web. Demand open, rights-respecting funding alternatives.
- Next steps: Audit your site’s AI agent access, strengthen privacy UX, and pilot subscriber/donation models with transparent data practices.
## 🧱 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 browser conditions access to news
> [!quote] Democracy and the open internet die in daylight (heatherburns.tech) by [[Heather Burns]] via [[heatherburns.tech]]. Published 2025-10-22. Accessed 2025-10-22.
> “You get free access to news if you download PerplexityAI’s proprietary browser.”
Two lines: example of gatekeeping, used to argue lock-in risk.
[view highlight](https://heatherburns.tech/2025/10/22/democracy-and-the-open-internet-die-in-daylight/#H1) #xref #articleseed ID:rw_[MISSING]
### [H2] Journalism shifts from adtech to agentic AI
> [!quote] Democracy and the open internet die in daylight
> “Day 0 of an experiment in shifting the funding model of journalism from adtech to agentic AI.”
Two lines: claim of emergent funding pivot; use for trend-mapping.
[view highlight](https://heatherburns.tech/2025/10/22/democracy-and-the-open-internet-die-in-daylight/#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 browser gatekeeping undermines open access
- Scope: when access requires proprietary, AI-centric browsers.
- Pointer: article statement on free access requiring Comet.
- Links: [[Open-Web]] [[Lock-In]]
#### [Z2] Agentic AI replicates adtech’s privacy harms
- Scope: AI-driven distribution deals tracking user behavior.
- Pointer: author’s “privacy-obliterating” claim.
- Links: [[Privacy]] [[Journalism-Funding]]
---
### ✏️ Fleeting Queue (Extract by Selection)
- [F1] Draft policy checklist for AI agent access controls — context/https://heatherburns.tech/2025/10/22/democracy-and-the-open-internet-die-in-daylight/ #fleeting #inbox
- [F2] Compare revenue outcomes: subs vs AI referrals for mid-tier sites — context/ibid #fleeting #inbox
- [F3] Create decision tree for funding models beyond adtech — context/ibid #fleeting #inbox
---
<!-- deep-read:sections end -->