The list below maps the AI-enabled surveillance threats catalogued in Anthropic's Detecting and countering misuse of AI: September 2026 report to lines from Judas Priest's "Electric Eye." The band was writing fiction. The report isn't.
1. Overhead & satellite reconnaissance
GTG-30005
"Up here in space / I'm looking down on you / My lasers trace everything you do"
What it is: Military surveillance and targeting software built by an Iranian threat group to track US naval forces.
How it works: Threat actors used AI to build a Python pipeline that automatically gathered commercial satellite imagery, tracked ship and aircraft transponder signals, scraped captions from public military photos, and scanned shipboard communications gear for security flaws.
Why it's dangerous: It turns scattered web data and satellite photos into active targeting handbooks — capable of tracking, and potentially aiding strikes against, naval ships and service members.
2. Population-scale mobile interception — Lakana 360
GTG-50027
"I'm elected electric spy... I keep the country clean / There is no true escape / I'm watching all the time"
What it is: A nationwide mass-spying platform named Lakana 360, engineered for Mali's state intelligence service (ANSE) by a consultant using AI.
How it works: The platform connects directly into all three national mobile networks to monitor roughly 25 million SIM cards across Mali. It records calls, reads texts, uses AI voiceprint recognition to track people even after they swap burner phones, and generates AI intelligence dossiers on any phone number — with court warrant requirements intentionally disabled.
Why it's dangerous: Malian security services have a documented history of detaining and abducting journalists, opposition leaders, and activists. Removing judicial oversight gives state police unchecked power to hunt down and arrest citizens.
3. State cyber police & dissident scouting
GTG-14021
"Always in focus / You can't feel my stare / I zoom into you / But you don't know I'm there"
What it is: Domestic sentiment tracking and international dissident monitoring run by Chinese municipal security organs and cyber police.
How it works: Automated AI workflows scanned domestic social media feeds, queried government databases, tracked targeted activist accounts, and collected pre-operational venue details — scouting protest routes and event halls for pro-democracy marches in cities including Vancouver, Turkey, and Oslo.
Why it's dangerous: It generates target lists for state police enforcement — including coercive "talk to" interrogations, blocking citizens from petitioning the government, and harassing overseas activists.
4. Secret browser spying & database profiling
GTG-34007
"You think you've private lives / Think nothing of the kind"
What it is: An Iranian state surveillance operation combining a secret browser extension with Arman, a central government surveillance database.
How it works: Threat actors created a malicious Firefox extension — disguised as a harmless prayer-time utility — to secretly harvest social media identities. They also used AI to build web interfaces for Arman, a master database holding citizens' national IDs, personal beliefs, criminal records, and social accounts.
Why it's dangerous: Everyday users downloading a basic browser add-on have their private social media accounts secretly tied to their national ID and logged into state police files.
5. Hidden keyloggers & screen captures — SECOMS64
GTG-30006
"My tearless retina takes pictures that can prove"
What it is: A stealthy Windows spying implant called SECOMS64, built using AI by an Iranian threat group.
How it works: The software hides on a victim's computer behind a fake Telegram app icon and process name. Once hidden, it logs every keystroke — especially while the victim is chatting in Telegram — takes secret screenshots, steals Chrome passwords, and grabs internal Microsoft 365 email.
Why it's dangerous: Victims have no idea their private photos, typed messages, passwords, and live screen displays are being continuously captured and sent to attacker-controlled Telegram channels.
6. Demographic profiling & sentiment scoring
GTG-54009
"I take a pride in probing all your secret moves / I feed upon your every thought"
What it is: A commercial surveillance platform built by an intelligence vendor (S2T Unlocking Cyberspace) to profile social media users in Iran and the Persian Gulf.
How it works: The platform feeds batches of social media posts into AI to sort users into six demographic segments — urban, youth, military, clerical, rural, diaspora — pinpoint their physical locations, rate their political leanings, and generate government-style intelligence briefs in formal Arabic.
Why it's dangerous: It lets governments or private buyers map entire populations, single out critics, and target specific demographic groups with propaganda or police action.
7. Dark-web citizen doxxing engine — fafsearch
GTG-50029
"Develop and expose... And so my power grows"
What it is: A compiled search engine called fafsearch, created by a single French-speaking hacktivist using AI coding tools.
How it works: The actor built a platform that ingested tens of millions of exfiltrated records — including national health IDs, justice system breach data, and hacked political donor lists — and hosted it as an anonymously accessible dark-web search service.
Why it's dangerous: Anyone on the dark web can search political figures, campaign donors, or party members by name and instantly expose their home addresses, health records, and political affiliations.
Seven operations, one report, one song written forty-four years before any of this was technically possible. The lyrics were metaphor. The threat model is now a checklist.