Opinion / Privacy
God’s Eye View Shows How Public Data Can Become a Surveillance Weapon
Public visibility should not give strangers unlimited power to watch, identify, and catalog ordinary lives.
You should not have to disappear from public life to keep strangers from building a dossier about you.
Walking down a street, attending a political meeting, visiting a doctor, or going to vote should not expose someone to continuous monitoring by anyone with a computer. The growing ability to combine public cameras, location information, and automated analysis threatens that freedom.
God’s Eye View makes this problem visible.
The open-source project brings aircraft, ships, satellites, public cameras, and other information into a three-dimensional globe. Its presentation borrows heavily from the look of an intelligence operations center. The documentation also acknowledges important limitations: traffic can be simulated, camera coverage estimated, and information delayed or incomplete. Its thermal and night-vision appearances are visual effects, not proof that an infrared sensor is observing the scene.1
One fact deserves explicit recognition: God’s Eye’s maintainers prohibit features for named-person searches, facial recognition, and tracking individuals. They say contributions crossing that boundary will not be accepted. That is a responsible restriction. An article criticizing the project must acknowledge it.2
My criticism concerns the broader product model: making scattered observations easier to collect, navigate, and combine, while presenting that power as entertainment. A developer’s contribution policy cannot control every downstream modification or every use of the underlying information.
The foreseeable consequences deserve more attention than the visual spectacle.
A public camera might have been installed to show road conditions. A public post might describe an event. A transportation feed might identify a vehicle’s reported position. Individually, those disclosures have limited meaning. Combining and retaining observations can create something much more intrusive: a system for reconstructing activities, inferring relationships, and directing attention toward particular people.
Aggregation changes the risk.
The distinction is familiar in ordinary life. Someone might notice you entering a building. You would feel very differently about a stranger maintaining a searchable record of your movements and associations.
The fact that an observation was possible does not settle whether automating, preserving, and distributing it should be acceptable.
The most dangerous potential uses are painfully ordinary.
An abusive former partner could use surveillance capabilities to undermine a victim’s efforts to escape contact. A harasser could use apparent location information to direct threats or unwanted confrontations. Someone seeking to intimidate a witness or journalist could exploit information suggesting where that person spends time.
These are foreseeable abuse scenarios involving surveillance systems and possible extensions—not findings that God’s Eye or its developers have committed those acts.
Political intimidation presents another serious risk. A system used to identify or catalog people around demonstrations, campaign events, or voting locations could make lawful participation feel dangerous. A voter should not have to wonder whether appearing in public will result in an online accusation, an employer complaint, or a threatening message.
Presence also proves far less than a surveillance enthusiast might assume. Being near a religious building does not establish someone’s religion. Being outside a clinic does not establish a medical condition. Being beside a political event does not establish support for its organizers.
A system can expose someone and misrepresent them at the same time.
The danger becomes greater when an operator adds identity matching. Facial recognition can attach a name to the wrong person, and people receiving an alert may act before questioning it.
That concern has documented counterparts. In its case against Rite Aid, the Federal Trade Commission alleged that the retailer’s facial-recognition system generated thousands of false-positive matches. According to the complaint, employees responding to erroneous alerts followed customers, searched them, expelled them, publicly accused them, and sometimes involved police. Those allegations concerned Rite Aid’s system, not God’s Eye. They show the human consequences that can follow a supposedly authoritative identification.3
Buying better cameras does not eliminate the problem. NIST explains that false-positive differences across demographic groups can occur even with high-quality photographs. Accuracy depends on the algorithm and circumstances; a clear image is not a guarantee of a correct identification.4
The interface itself deserves scrutiny. A polished map, tracking box, or intelligence-style display can encourage confidence that the underlying evidence does not justify. An estimate can look like an observation. A stale position can look current. An association can look like proof.
For the person wrongly identified or publicly accused, an explanation buried in documentation may arrive far too late.
There is also a basic accountability problem. The person exposed by a surveillance system may have no relationship with its operator, no notice that monitoring is occurring, and no practical way to demand correction or deletion.
“Publicly available” is an inadequate answer to those concerns. It describes accessibility. It does not establish meaningful consent to every subsequent use. Even God’s Eye’s licensing documentation distinguishes its open-source code from third-party data, which remains subject to separate terms.5
Regulators have already recognized the dangers of sensitive location information in another setting. In 2024, the FTC finalized restrictions on data broker X-Mode and its successor Outlogic concerning sensitive location data. The order addressed information capable of exposing visits to places such as medical facilities and houses of worship, and included safeguards concerning demonstrations and identifying or locating specific individuals. That case involved brokered location data rather than public camera feeds, but the underlying concern is directly relevant: location information can expose intimate activities and create opportunities for abuse.6
We need enforceable limits that address the harm before someone becomes a victim.
Public agencies should have to justify publishing raw, identifiable imagery. Where congestion measurements, delayed images, or appropriately obscured footage serve the public purpose, unrestricted identifiable video should not be the default. Retention should be limited, and sensitive locations should receive additional protection.
Legislatures should prohibit indiscriminate biometric identification and persistent person tracking through public feeds, with narrowly defined exceptions subject to meaningful oversight. Government access should require appropriate legal authority and accountability.
Operators should face obligations concerning collection, retention, security, and downstream use. People harmed by unlawful surveillance should have practical ways to obtain correction, deletion, injunctive relief, and compensation.
These rules should preserve journalism, government transparency, lawful public recording, and consensual research. Public spending, official conduct, and institutional abuses must remain open to scrutiny. Protecting those functions does not require turning ordinary people’s daily lives into searchable surveillance material.
God’s Eye’s refusal to add individual tracking is a useful boundary. The larger ecosystem needs boundaries that survive beyond one project’s maintainers.
The question we should ask of an impressive surveillance interface is simple: What power does this give an abusive operator over someone who never agreed to be watched?
If the answer includes persistent identification, intimidation, or the exposure of sensitive activities, unrestricted access should not be treated as an achievement.
We should build public information systems around a clear principle: people retain a right to live ordinary lives without becoming targets.
Evidence & attribution
Read the original sources.
Project documentation describes capabilities and limitations. The FTC and NIST sources concern separate systems; they do not establish misconduct by God’s Eye View.
- God’s Eye View: official project documentation
Public data layers, modeled traffic, visual styles, and stated limitations.
Return to text ↑ - God’s Eye View: Responsible & Open
The maintainers’ prohibition on named-person searches, face recognition, and tracking individuals.
Return to text ↑ - FTC: Rite Aid facial-recognition allegations and proposed restrictions
December 19, 2023. The article attributes the described incidents and system failures to the FTC’s complaint.
Return to text ↑ - NIST: Demographic Effects in Face Recognition
Evaluation findings concerning false-positive and false-negative differences, image quality, and algorithm variation.
Return to text ↑ - God’s Eye View: Data Sources & Attribution
The distinction between the project’s source-code license and third-party data and asset terms.
Return to text ↑ - FTC: Final X-Mode / Outlogic sensitive-location-data order
April 12, 2024. Restrictions concerning a separate data broker and its handling of sensitive location information.
Return to text ↑