Tools & Resources

Extract text from scanned documents, images, and PDFs with open-source OCR engines. Essential for archival research, leaked document analysis, and historical record digitization.

Workflow Example: Download leaked PDFs → OCR with DeepSeek/PaddleOCR → Extract entities (names, amounts, dates) → Store in SQLite FTS5 → Search and cross-reference → Export findings as structured CSV/JSON.

Build Your Own Solutions

Instead of relying on corporate data services, use free and open-source APIs to collect, analyze, and visualize data directly. These resources are maintained by communities and trusted organizations.

Corporate platforms are losing their value. Here's why your alternative matters.

The problem: For over a decade, corporate digital services relied on a closed model: proprietary platforms, opaque pricing, licensing lock-in, and centralized control. This worked because there were no credible alternatives at scale.

That condition no longer exists. Large language models (LLMs) combined with mature open-source ecosystems are structurally undermining the value proposition of big tech products.

What this means for you:

• Evaluate alternatives: Before buying a corporate tool, check if an open-source equivalent exists.
• Build your own stack: Combine specialized open tools instead of relying on one corporate platform.
• Support interoperable systems: Prefer tools that export data, use open standards, and don't lock you in.
• Join communities, not platforms: Open-source projects are controlled by users, not shareholders.

Why This Matters

In 2025, over 60,000 pages of Epstein-related documents were officially released by the U.S. House Oversight Committee and federal courts. These documents expose institutional failures, complicity networks, and power dynamics that enabled decades of abuse.

📊 Complete Analysis Guide

Download the comprehensive guide: document acquisition workflow, full-text search database setup, entity extraction, co-occurrence analysis, and reproducible methodology for investigating power networks.

Includes: JMail.world email database, text extraction pipeline (Python + OCR), SQLite FTS5 database, entity mapping, co-occurrence analysis, timeline reconstruction.

Why reclaim digital infrastructure?

• Corporate control = data weaponization: Google, Amazon, Meta, Apple track everything. Your data is sold to state agencies (NSA, GCHQ, law enforcement).
• Algorithmic manipulation: Corporate platforms optimize for engagement, polarization, and compliance. Your feed is a behavioral control tool.
• Arbitrary enforcement: Sudden bans, suspensions, content removal without due process. No appeal. No transparency.

🌐 Browser & Search — Reclaim Privacy

• Chrome (Google tracking) → Brave (built-in ad-blocking), Firefox (open-source), Ungoogled Chromium (privacy fork).
• Google Search → DuckDuckGo (no IP logging), SearXNG (privacy metasearch).
• YouTube → Invidious (YouTube without tracking/algorithm).

💬 Social Media & Communication

• Instagram, TikTok → Mastodon (federated, no algorithm), Pixelfed (photo sharing).
• Twitter/X → Mastodon (open, decentralized).
• Facebook → Diaspora (federated), Signal (encrypted messaging).

🎬 Media Centers

• Netflix, Disney+ → Jellyfin (self-hosted media), Kodi (media center).
• Spotify → Subsonic, Airsonic (self-hosted music).

All of these are legal, production-ready, and often superior to corporate alternatives. You keep your data. No account bans. No algorithm.

How to use templates in 10 minutes

1. Open the most relevant template (petition, mp-email, briefing).
2. Edit the first line to add target, dates and a short ask—avoid jargon.
3. Add 1–2 evidence lines and link to the briefing or monitoring log.
4. Run OSINT safety checklist: citations, image provenance, doxxing checks.
5. Publish, then log the link and key metrics (signatures, replies, mentions).

Quick MP message example (copy-and-paste):

Dear [MP Name],
I'm a constituent and I urge you to request an independent audit of contract [#] due to concerns about [brief impact]. Please ask the minister to pause any further disbursements until the review is complete.

See the briefing template for evidence structure. When in doubt, consult the OSINT safety resources before publishing.

Toolkit: Investigation, Whistleblowing & Real Impact

• OSINT & Investigative Journalism: Bellingcat How-To, Lighthouse Reports, OCCRP for image verification, arms tracking, satellite analysis, collaborative investigations.
• Secure Whistleblowing: GlobaLeaks, DDoSecrets, SecureDrop for anonymous reporting of abuse and corruption.
• Scraping & Data Leaks: Open-source tools for extracting procurement, lobbying, arms export data. Cross-check with LobbyFacts and Transparency International.
• Legal Support & International Complaints: FIDH, Amnesty International, Transparency International for templates to file with ECHR, UN, and free legal aid.
• Public Dashboards & Campaigns: Datawrapper, Flourish, Google Sheets for publishing findings and building public pressure.

⚙️ Practical AI Workflows

Don't let AI do your thinking for you. Use it to save time on boring stuff so you can focus on what matters: understanding your issue, talking to people, organizing action.

📖 Read the full resources → AI for Campaigns: A Practical Resources for Activists

• Summarizing reports: Feed a 100-page UN report to an LLM, get a 1-page summary with key facts. Then verify it against the original.
• Finding patterns: Train a model on news articles to detect when escalation language appears. Helps you spot when things are getting worse.
• Cleaning messy data: You have a spreadsheet of corporate contracts in 5 different formats? Use data tools to standardize and cross-check.
• Generating outreach: AI can help personalize emails to MPs at scale. But YOU write the message, YOU verify the facts.
• Organizing evidence: Use version-controlled databases to track who said what, when. Tamper-proof record keeping.

Safety & Ethics: Using AI Responsibly

Never trust AI alone. It's a tool. It makes mistakes. It can hallucinate. It can be biased.

• ✓ Verify every claim against primary sources
• ✓ Use local models in high-risk countries (no cloud API logging)
• ✓ Keep datasets encrypted and version-controlled
• ✓ Train models only on verified data
• ✗ Never expose sources in plaintext to commercial LLM APIs

See Privacy & Security resources for threat modeling and safe AI use in restricted contexts.