Ex-Military Hacker: The Secret World Of Government Surveillance - Bill Thompson
In a Nutshell
Ex-military hacker Bill Thompson explains how the Patriot Act and FISA court enable broad government surveillance with minimal oversight, while intelligence agencies recruit insiders using placement, access, and motivation criteria. He reveals that human intelligence operations cost 95% less than technical exploits like Pegasus spyware, and that AI now reduces target development time from months to hours by analyzing device configurations. Thompson advocates cutting federal government by 80% to eliminate waste, noting that agencies manufacture threats to justify budgets while operating outdated systems that barely function.
These notes were generated by AI and may contain inaccuracies.
Bill Thompson uses a Galaxy Fold rather than an Apple device due to a philosophical preference for open systems. The Android Open Source Project (AOSP) allows visibility into the phone's inner workings, enabling users to identify vulnerabilities and perform their own forensic analysis. This capability proved valuable during military work developing implants and zero days, as well as when traveling to countries where phone security might be compromised or when phones might be seized at customs.
Apple's closed architecture makes such self-auditing impossible, though proponents argue this opacity provides security through obscurity. Thompson prioritizes the ability to personally examine his device over trusting corporate security claims, weighing Apple's trustworthiness against the competence of those analyzing Android vulnerabilities.
The Patriot Act requires minimal justification for accessing phone data, having been written to target telecommunications infrastructure including microwave links and base transceiver stations (BTS). The legislation's broad definition of "infrastructure" effectively future-proofed it for emerging technologies. Thompson describes the Patriot Act as one of the most usurpacious pieces of legislation, noting the ironic naming convention where innocuous terms mask power-grabbing measures.
The Foreign Intelligence Surveillance Act (FISA) court operates as a secret judicial body near Fort Meade, Maryland, serving the NSA and other intelligence agencies. When targeting individuals, agents present evidence to FISA judges in closed proceedings that never become public. Thompson notes this system extends to major technology platforms including Meta and OpenAI, where government requests for user data like ChatGPT logs likely occur through both formal memoranda of understanding and clandestine recruitment of insiders.
Intelligence agencies use a three-part framework for recruiting sources: placement (does the person work in the target organization), access (do they have the desired information), and motivation/susceptibility (can they be motivated to share it). Thompson was trained as a human intelligence operator in the military, learning to recruit individuals to compromise their own countries' security.
When direct recruitment proves difficult due to legal constraints, agencies may use Five Eyes partnerships—American agencies might know British intelligence has recruited a source they cannot officially target, then receive information through partner debriefings.
Minimization procedures are designed to protect American citizens and dual citizens from unlawful surveillance under the Bill of Rights. When inadvertent collection occurs—such as an American appearing on a monitored cell phone channel in Bahrain—the government is supposed to purge this information. However, Thompson describes these processes as moving "conveniently slow," creating another vector for rights violations. Figures like Rand Paul have publicly opposed FISA and minimization procedures.
Technology development outpaces legislative response, with government always lagging behind those building new systems. DARPA represents one of the few competent government organizations, functioning as the Department of Defense's "black box" for developing unconventional ideas. DARPA funds speculative projects knowing many will fail but hoping for valuable spin-offs—like developing advanced carbon fiber while attempting to build faster spacecraft.
Human intelligence involves recruiting sources, while technical operations focus on exploiting computer systems and developing zero-day vulnerabilities. Thompson's first post-military job offer came from a bank seeking penetration testing of their own infrastructure—a "victimless ransom" that strengthens security.
The choice between human and technical approaches depends on budget, infrastructure fortification, and operational security practices. Human recruitment typically costs 95% less than technical exploitation, though this varies by target technology. A 3G Nokia in Iran represents an easy target, while highly encrypted satellite networks requiring physical positioning within satellite coverage areas demand extensive resources.
Traditional human intelligence operations involve identifying targets with exploitable vulnerabilities—such as financial needs for medical procedures like eye surgery or cancer treatment—and building rapport through seemingly natural encounters. This "bump" recruitment process establishes trust before transitioning to paid information sharing.
Forensic exploitation bridges human and technical intelligence, involving GPS and phone data analysis to verify whether recruited sources are following operational instructions. This includes checking if assets conduct proper surveillance detection routes rather than taking direct paths that could expose operations.
Pegasus represents one of the most sophisticated publicly known exploits—a persistent implant that originally required click-based delivery through phishing messages appearing to come from trusted contacts. The malware installs in phone locations that evade standard forensic triage, making detection difficult even for trained operators. The Israeli-developed tool has since evolved beyond its initial iOS-specific implementation.
Thompson discusses sophisticated supply chain interdiction operations involving the compromise of pager devices, noting the extensive planning required including corporate espionage, manufacturing infiltration, and patient timing. Such operations demonstrate how adversaries adapt to smartphone insecurity by reverting to older, supposedly more secure technologies.
Throughout the global war on terrorism, opposing forces demonstrated rapid technology pivoting—moving from pressure-triggered IEDs to command-detonated devices, then to low-frequency garage door openers when initial RF detection systems proved effective. This cat-and-mouse dynamic shows how asymmetric warfare enables rapid adaptation outside Geneva Convention constraints, similar to drone warfare innovations observed in Ukraine.
Modern warfare involves rapidly evolving counter-drone tactics. Tanks have been fitted with prongs at the top to catch drone wires. EMP blasts are used to knock drones out of the sky. Large strings and cords are deployed as countermeasures. The military and government struggle with pivoting and responding to rapidly changing threats because they're built around symmetric threats like Russia and China, operating under Geneva Conventions and laws of land warfare with gentleman rules about not targeting civilians, churches, or mosques.
Terrorists exploit these self-imposed restrictions by targeting hospitals and using them as torture facilities. Intelligence failures are documented in popular history books, with one such book found on Osama bin Laden's bedside table after the raid.
Bill Thompson advocates reducing the federal government by 80%, citing numerous organizations that should be eliminated. Government operates as a self-licking ice cream cone focused on growth and pulling more purse strings. The top 1% of earners already pay 30% of taxes, but the primary concern among high-taxpaying individuals is waste and fraud rather than tax rates. DOGE's focus on removing inefficiencies represents an attempt to make government systems more efficient rather than simply increasing tax revenue.
Sam Caucos, CIO for the Treasury Department and founder of Levels, described government systems as ancient and non-agile. The organization is a lumbering behemoth that leans left internally and cannot keep systems updated. Treasury operations involve 16 intersecting different systems requiring cross-database lookups rather than a single coherent system. These Frankenstein-like systems function only to the extent that the country hasn't fallen apart and taxes can still be submitted and paid.
Bill Thompson's last military job involved advising a two-star general and full colonel in charge of all offensive cyber development for the Army, known as OFFCO. Budget execution drives all operations—if a general receives $100 billion and spends only $99 billion, they face severe consequences. The National Intelligence Priority Framework (NIPF) outlines effects the military needs to achieve and identifies peer competitors and countries to avoid targeting.
Budget and money drive everything in government operations. Meeting the president's requirements matters less than executing the full budget, even if this requires lying about threshold achievements. When money remains unspent by July with a September budget deadline, military personnel are pressured to find ways to spend remaining funds immediately, regardless of operational value. A general who achieves presidential effects with only 70% of the budget should receive promotion rather than criticism.
AI code generation is robust but not as advanced as commonly believed. It excels at generating mundane, repetitive code quickly, reducing development time from weeks to hours. In hacking applications, AI can analyze technical dumps of target devices including routers, examining IP table configurations, firmware, software, and network diagrams to identify vulnerabilities. What previously took 3-6 months or 1-2 years for target development now takes a month at most, with most work completed in the first 3 hours.
AI development depends on training data and compute power. The mathematical foundations including convolutional neural networks and long short-term memory neural networks existed since the 1970s-80s, but lacked sufficient compute (now provided by GPUs and tensor processing units) and training data (now provided by the internet). Organized crime groups are training their own models on private code repositories using personal GPU setups without guardrails.
Organized crime groups primarily use hacking for phone access, banking operations, Bitcoin-related activities, and betting operations.
Human intelligence operations require skills in car pickups, car drop offs, bumps, initiating contact, debriefing, and source handling. Source categories range from Cat 4 (lowest value, such as gate guards who can provide access cards) to Cat 1 (highest value, such as recruiting a Secretary of Defense). Most people can recruit lower-category sources using structured military approaches, but Cat 1 sources require artisans who understand target psychology, motivation assessment, and operational security.
Elite human intelligence operators combine rigorous methodology (thema) with artistic adaptability (animus) to handle chaotic situations. They can quickly assess individuals in 15-minute conversations, understanding motivations, tactics, and maintaining operational security. Most reach this level of expertise around age 40-45. Three such operators could theoretically take over a country.
Increased digital tracking through forensics, DNA, fingerprinting, and digital footprints has made traditional intelligence operations more difficult. Handling Cat 1 sources has become more restricted, requiring different cover approaches. Instead of casual bar meetings, operations now involve setting up legitimate companies in target sectors, with governments spending significant money to establish CEO positions for intelligence operatives. This shifts from turning existing people to creating new cover identities that can approach high-value targets through business relationships.
Chinese intelligence monitors critical supply chain partners from the outset of any business relationship. Before engaging in buying or selling, they conduct thorough surveillance to understand all operational details. When a meeting occurs, representatives appear in person to inquire about relationships with specific individuals, demanding that these connections be severed.
Technical collection capabilities cannot be altered, so alternative approaches involve establishing competitive or adjacent technology companies fully funded by the US government through cutouts. These operations place individuals in key positions such as chief of staff to create products that are slightly cheaper than existing options. This provides cover for action and cover for status, allowing legitimate business explanations when Chinese representatives inquire about supplier relationships.
Elon Musk ranks among the most surveilled humans on the planet. Recruitment targets include individuals close to high-profile figures rather than the principals themselves due to accessibility constraints. Family members of Trump would also be recruitment targets depending on the recruiting nation.
Accessibility serves as a critical vector in target recruitment. Even with personal connections, high-profile individuals remain inaccessible for debriefings or information gathering. Recruitment focuses on people surrounding the target rather than the target directly.
Clinton would represent a nightmare recruitment target due to management difficulties and high visibility. The argument against Israeli intelligence recruiting Epstein centers on his obvious nature as a target - being very flash, lavish, and public-facing makes him an unsophisticated choice for intelligence operations.
Visible individuals face increased accusations of being controlled opposition or federal agents. The visibility-to-glow ratio suggests highly visible people make poor recruitment candidates because they attract hundreds of thousands of daily observers.
The federal government funds both sides of conflicts to maintain enemies and straw men for operational purposes. This includes KKK chat rooms and January 6th events where federal agents were present in audiences. Historical precedent includes the Gulf of Tonkin incident where government involvement created motivation for military operations.
The FBI maintains divisions focused on threats to the homeland with 25% of budgets allocated to white supremacy. When actual threats prove minimal, agencies create the conditions necessary to justify continued funding and job preservation. This creates self-licking ice cream cones where organizations manufacture the problems they were designed to counter.
Big Pharma profits from ongoing medication rather than cures, similar to government agencies that perpetuate problems to maintain budgets. The alternative approach involves working within existing systems to achieve desired outcomes while maintaining profitability, rather than attempting systemic overthrow.
Excessive information creates confusion and apathy rather than certainty. The desired outcome may be deliberate confusion that prevents conviction in any direction. This meta-level analysis suggests information flooding as a strategy to neutralize opposition.
Government involvement extends to approving toilets and light bulbs, with taxation occurring at every stage of earning, spending, saving, and death. The founders intended constrained government with paramount individual rights, but current scope exceeds constitutional bounds to the point where founders would advocate for revolutionary response.
North Dakota presented limited opportunities beyond farming or meth addiction support industries. Military recruitment offered escape from trailer home life, family addiction issues, and lack of purpose. Initial enlistment focused on personal escape, while re-enlistment centered on preserving the American experiment after studying Montesquieu, John Locke, Cicero, and the Federalist Papers.
The American system enables the greatest economy, technology, culture, and wisdom in history. The decision to bet on individuals, families, and freedom of choice stems from belief that people make better decisions when left to their own devices than any government could provide.
Spartan Forge addresses post-military transition challenges by applying military target analysis skills to hunting scenarios. The app uses neural networks trained on GPS collar data from deer to predict movement patterns. High-resolution 5cm imagery exceeds Google Maps quality and includes historical data access through contracted aerial photography partnerships.
Rather than competing in the mapping space dominated by established corporations, the focus remains on machine learning and neural networks for geospatial intelligence applications. The technology serves hunters with limited time availability and addresses broader needs in soil analysis, mineral rights, real estate, and solar panel optimization.
The platform integrates high-resolution imagery collected over time with additional data sources including LAR data and other inputs. This combination enables visual understanding of terrain by removing tree cover through light detection methods similar to radar technology. The data can be converted into 3D representations, allowing users to see paths and features beneath the surface.
During the hurricane in North Carolina, first responders received accounts to access the system. When roads were washed out and houses were inaccessible, users identified historical logging trails through the app to reach isolated homes and deliver critical supplies including diabetes medication and oxygen for emphysema patients.
The blue force tracker capability allows users to set pins marking washed-out roads and other hazards. This same functionality serves hunters for organizing group activities by dropping pins that all participants can see, providing shared access to high-resolution imagery, LAR data, historical imagery, and neural network analysis.
The company developed its own neural network called Cyber Scout specifically trained on hunting data. Unlike general AI systems that regress to the mean by pulling from broad internet sources, Cyber Scout uses retrieval documents sourced exclusively from the highest caliber hunting information. This focused training enables superior guidance for tasks like bow tuning, walking users through processes more effectively than general chat systems.
Cyber Scout will soon integrate with mapping features to provide area-specific hunting guidance and strategic recommendations.
The company launched in 2017, released its first mobile app in 2021, and currently ranks around number 14 in mapping and outdoors categories in the app store with hundreds of thousands of users.
The foundational principles emphasize the importance of tribe, friends, wilderness connection, and obtaining resources. The app addresses modern constraints where people have limited vacation time and may need to scout public land in distant locations. Users can remotely analyze terrain using high-resolution imagery and other tools, then download offline maps for actual field use.
Notable user experiences include fathers and sons successfully harvesting deer after years of targeting specific animals, using the neural network predictions and high-quality mapping to identify productive locations. The platform enables users to drop pins marking hunting locations for family safety, allowing offline tracking if someone becomes injured and needs rescue assistance.
Thompson describes his military role as building advanced technology for special operations personnel rather than direct combat involvement. He characterizes himself as "the girthy part of the staff that was supporting the head of the spear" rather than operating at the tip of operations.
Users can access the platform at SpartanForge.ai or through the app store. Thompson personally manages social media interactions, spending approximately two hours each morning responding to user questions, complaints, and feedback. Technical issues get escalated to the QA team, but general inquiries receive direct responses from the product owner.
The episode concludes with recognition of partners who made a gift possible, including Hoy, Archery Country, and Dialed Archery. Specific individuals mentioned include Tom from Hoy, Tyler from Archery Country, and Taylor and Jordan from Dialed Archery, noted as American-made companies that Thompson works with regularly.
Keep Chris Williamson in your library
Save the videos and channels worth coming back to, and find them again in one place.
