AI-ASSISTED OSINT

Get oriented with the course structure, meet Vlad Sutea, and set clear expectations for the seven weeks of AI-assisted OSINT ahead.

• Instructor introduction
• Course premise and philosophy
• Course structure and tool setup
• Assignments and final project overview

Differentiate intelligence from raw information, and evaluate how AI is reshaping OSINT workflows while examining the ethics and legal boundaries that keep investigations defensible.

• Intelligence vs. information
• AI's role in OSINT: help vs. hype
• Case study: AI-assisted satellite imagery on Russian tank losses
• Case study: BBC Africa Eye's Anatomy of a Killing
• Ethics, OPSEC and the CFAA
• Evidence standards for defensible investigations

Compare leading LLMs for OSINT tasks, learn prompt engineering fundamentals for aggregation and synthesis, and build a reusable prompt library you'll use throughout the course.

• Claude, ChatGPT, Gemini and open models compared
• Trusting vs. verifying LLM output
• Prompt engineering fundamentals
• Building reusable prompt templates
• Demo: Building an AI-Assisted Research Workspace

Assignment #1: Build Your OSINT Prompt Library
Create a reusable investigation workspace with one project context file and two prompt templates you'll reuse throughout the course.

Set up a secure investigation environment and learn what should never be pasted, uploaded or connected to an AI tool during live work.

• Managed identities, dedicated browsers and VPNs
• Visibility analysis of your own digital footprint
• Safe AI use during investigations
• Risks of model memory and excessive agent permissions
• Workshop: Personal Digital Footprint & AI Safety Assessment

Explore what AI agents actually do inside an OSINT workflow, and learn no-code ways to automate monitoring and collection safely.

• What AI agents are, and aren't
• No-code automation using Claude
• Safe, legal automation at scale
• Guest speaker: Max Blumenfeld on AI agents for OSINT

Assignment #2: Design and Test an OSINT Agent Workflow
Design a non-technical blueprint for an AI agent supporting a recurring OSINT task, then tabletop-test whether it's useful, safe and realistic.

Learn to spot deepfakes, voice clones and AI-generated images, then build a defensible verdict using metadata, reverse image search and provenance analysis.

• Metadata, reverse image search and InVID
• The synthetic media landscape: deepfakes, clones, generated text
• Building an authenticated, debunked or uncertain verdict
• Workshop: Detecting & Debunking Synthetic Media

Master geolocation and chronolocation tradecraft, matching imagery to place and time using satellite sources, shadows and sun position.

• Geolocation and chronolocation fundamentals
• Demo: Comparing LLMs for Geolocation
• Workshop: Geolocating an Open-Source Investigation
• Google Earth Pro, SunCalc and reverse image search

Assignment #3: Media Verification & Evidence Challenge
Complete a verification challenge with AI assistance and document a defensible evidence chain showing how you reached your conclusion.

Interpret publicly available satellite imagery using AI-assisted techniques, covering change detection, entity recognition and the limits of resolution and confidence.

• Open-source GEOINT fundamentals
• Change detection and entity recognition
• Demo: AI-Assisted Satellite Imagery Interpretation
• Google Earth Web, Copernicus and FIRMS

Turn unstructured social media chatter into structured intelligence, then map the networks and relationships behind a real public claim.

• SOCMINT across X, Telegram, Facebook and TikTok
• Building research accounts safely
• Node-and-edge network analysis
• Demo: AI-Assisted Social Media Collection & Network Visualization

Assignment #4: Social Media Intelligence and Network Mapping
Investigate the network surrounding a public claim, organization or piece of media and build an evidence-based network map.

Assess the information environment for threat signals and narrative manipulation, drawing on documented cases of AI-powered influence operations.

• AI-enabled threat and disinformation strategies
• Sentiment and narrative mapping tools
• Case study: Doppelgänger influence operation
• Case study: Operation Overload / Matryoshka

Move from collection to analysis using Analysis of Competing Hypotheses, Key Assumptions Check and Red Teaming, with AI as a partner, not a crutch.

• Analysis of Competing Hypotheses (ACH)
• Key Assumptions Check and Devil's Advocacy
• Guarding against bias, yours and the model's
• Demo: Applying an LLM to Structured Analytical Techniques

Assignment #5: Narrative and Threat Assessment
Analyze narratives and threat indicators in a defined body of public information and apply a structured analytical technique to produce a reasoned assessment.

Evaluate what deep and dark web intelligence is actually worth, and where legal, ethical and OPSEC boundaries stop direct student access.

• Surface, deep and dark web defined
• Legal and OPSEC boundaries
• Third-party dark web monitoring signals
• Interpreting sanitized dark web intelligence reports

Learn US IC 203 analytic standards and estimative language, then use LLMs to draft executive intelligence products without losing analytical integrity.

• US IC 203 standards and Admiralty Code
• SITREPs, PDBs and Strategic Assessments
• Translating findings for the right audience
• Demo: Writing Executive Intelligence Products with LLMs

Map real OSINT job titles across public and private sectors, and build a sanitized, legally compliant portfolio for interviews and technical assessments.

• Public vs. private sector OSINT roles
• Portfolio building and certifications
• Interview prep for intelligence roles
• Guest speaker: Career pathways in AI-assisted OSINT

Assignment #6: Your OSINT Career Route Map
Identify an OSINT-related role you want to pursue and develop a realistic, personalized pathway toward it.

Present and defend your Executive Intelligence One-Pager to a professional audience, then get peer and instructor feedback on your final product.

• Capstone presentations by career track
• Verification, estimate, entity report or threat assessment formats
• Peer and instructor feedback