The scale of the system was detailed this week in an investigation by The Daily Aztec, SDSU’s independent student newspaper, which obtained university records showing the broad reach of the campus surveillance network. The reporting builds on an earlier Daily Aztec investigation revealing that SDSU’s University Police Department spent more than $1.3 million upgrading its camera system with technology capable of artificial intelligence-assisted detection, analysis and search functions.
According to the report, the university says the cameras are not being used for facial recognition, behavioral tracking or profiling. SDSU Police have described the current use as limited to basic motion detection, system maintenance and alerts in restricted areas outside normal operating hours.
But the controversy is not simply about what SDSU says it is doing today. It is about what the system can do, who controls it, what students were told, and whether a public university should install a campuswide AI-enabled surveillance infrastructure before adopting a new, specific policy governing the technology’s most sensitive capabilities.
According to The Daily Aztec, the cameras are installed in classroom hallways, entryways, bookstores, dining facilities, parking structures, recreation areas and residence halls. The student journalists found that more than 330 cameras were listed in residential halls alone, representing roughly 28 percent of the campus camera network. Huaxyacac, SDSU’s largest first-year residence hall, reportedly had 79 cameras installed, followed by Tenochca with 36 and Chapultepec with 33. The Daily Aztec also reported that 18 of SDSU’s 24 residential buildings appeared on university camera location records obtained through public records requests.
That matters because residence halls are not ordinary public spaces. While hallways, elevators, lobbies and common areas may not carry the same privacy expectations as a dorm room, they are still part of students’ living environment. Students shower, sleep, receive guests, attend meetings, return from medical appointments, participate in political organizing, and move through those spaces at all hours of the day and night. A camera in a residence hall hallway is not the same as a camera at a parking lot exit.
The California State University systemwide Video Security Cameras Policy allows cameras in public areas and recognizes that recordings must be made with respect for the reasonable expectation of privacy among members of the university community. The policy defines public areas broadly to include parking lots, hallways, library study rooms, buildings open to the general public and outdoor spaces. It prohibits cameras from viewing areas where there is a reasonable expectation of privacy and says cameras may not be directed or zoomed into the windows of private residential buildings, including residence halls.
The policy also requires notice that video cameras are present at campus entrances and at the entrance to any specific public area monitored by a camera where there may otherwise be a reasonable expectation of privacy.
SDSU told The Daily Aztec that students are informed about cameras through the Guide to Community Living and housing website. However, the student newspaper reported that the housing license agreements themselves do not mention the cameras. The Daily Aztec also found that the materials cited by SDSU do not specifically disclose the AI technology or its potential capabilities. That distinction is central to the dispute.
Traditional security cameras record what happens. AI-enabled camera systems can do more. Depending on the software activated and permissions granted, modern systems can classify objects, detect unusual activity, identify vehicles, search footage based on a person’s appearance, count crowds, flag motion in restricted areas, read license plates, detect sounds, integrate with other databases and generate alerts.
SDSU’s vendor, Avigilon, a Motorola Solutions company, markets AI video analytics as a way to detect anomalies, automate investigations and help security personnel respond more quickly. Avigilon documentation also describes facial recognition lists and related tools that can allow authorized users to create profiles and search for faces across enabled cameras.
SDSU says it is not using facial recognition or behavioral profiling. That statement is important, but it does not eliminate the policy question. Once the infrastructure exists, future use can change through software settings, administrative decisions, policy revisions, budget pressure, law enforcement demands or leadership turnover. Privacy advocates often call this “mission creep,” the process by which a technology introduced for one limited purpose gradually expands into broader uses.
The Daily Aztec’s earlier reporting found that SDSU documents described desired capabilities beyond simple maintenance alerts, including AI-based anomaly detection, license plate readers, gunshot detectors and the ability to search video footage for physical characteristics. University police have since said those more sensitive capabilities are not being accessed.
That leaves SDSU in a familiar position for institutions adopting surveillance technology: asking the public to trust that powerful tools will remain unused because current administrators say they will not activate them.
San Diego has already been through this debate. The city became a national example of surveillance backlash after thousands of “smart streetlights” were installed beginning in 2016 under a program initially presented as a tool for traffic, parking and environmental data. Residents later learned the streetlights also contained cameras that police could access for investigations. During the 2020 Black Lives Matter protests, San Diego police used the streetlight camera system dozens of times while investigating alleged vandalism, looting and property destruction.
The backlash helped fuel the city’s Transparent and Responsible Use of Surveillance Technology Ordinance, known as the TRUST Ordinance, which requires city departments to disclose surveillance technologies, publish impact reports, submit use policies and obtain public approval. The ordinance also created the city’s Privacy Advisory Board.
Even with those safeguards, San Diego’s surveillance fights have not stopped. The city’s use of Flock Safety automated license plate readers and smart streetlight technology has continued to face opposition from privacy advocates, immigrant rights organizations and civil liberties groups. The Electronic Frontier Foundation and local TRUST coalition members have warned that license plate reader networks collect information on ordinary people’s movements, can be shared improperly, and can chill First Amendment activity.
San Diego’s citywide fight is not identical to SDSU’s campus camera system. SDSU is governed by CSU policy, not the city’s TRUST Ordinance. But the underlying issue is similar: powerful surveillance technologies are often installed before communities fully understand how they work, how data is retained, who can access it, whether third parties can obtain it, and what remedies exist if the technology is misused.
For students, the stakes are unusually personal. A public university is not just a workplace or a municipal street grid. It is a place where students live, study, socialize, protest, seek medical help, attend religious and cultural events, meet with counselors, join advocacy groups and move through formative years of adult life. A camera network spread through campus can capture not only safety incidents but also daily patterns of association and behavior.
That raises First Amendment concerns. Courts have long recognized that surveillance of political activity can chill speech and association, even when no one is arrested. On a college campus, where protests, student organizing and controversial speech are core parts of public life, students may reasonably wonder whether attending a demonstration, meeting with an activist group, participating in a labor action or joining a contentious campus debate could place them in searchable video records.
It also raises Fourth Amendment questions, though those are more complicated. In general, people have reduced privacy expectations in public spaces. Cameras in hallways, plazas and parking lots are not automatically unconstitutional. But AI changes the practical analysis. The constitutional issue is not only whether a person can be seen in public. It is whether the government can use technology to aggregate thousands of moments into a searchable record of a person’s movements, associations and habits. That is the same concern now being litigated and debated nationally around automated license plate readers, facial recognition, drone surveillance and other AI-assisted policing tools.
California law also complicates the picture. California is a two-party consent state for confidential audio communications, and CSU policy says video security cameras installed under the systemwide policy may not capture audio. If AI-enabled cameras include audio detection capabilities as a technical feature, SDSU would need to ensure those features are not used in a way that violates state law or CSU policy.
Student records law may also come into play. Under federal FERPA guidance, a school surveillance video can become an education record when it is directly related to a student and maintained by the institution, such as when footage is used in student discipline. CSU’s own policy acknowledges that video camera recordings involving specific students are generally considered law enforcement records unless the university uses them for student discipline or makes them part of an education record. That means the same footage could occupy different legal categories depending on how SDSU uses it.
The campus labor implications are also significant. CSU policy states that video security cameras shall not be used for employee supervision, timekeeping, attendance, evaluation or general monitoring of employee conduct. It allows footage to be used to corroborate, prove or disprove specific acts of misconduct under certain circumstances, but prohibits routine use as a workplace surveillance tool.
That is an important safeguard, but it depends on documentation, access controls and enforcement. A policy that says cameras will not be used for employee supervision is only as strong as the logs, audits and consequences that accompany it.
SDSU’s situation also comes during a broader national expansion of AI-enabled security on college campuses. Michigan State University has pursued a system designed to detect barrier breaches, track individuals, count crowd sizes and read license plates. Other campuses and school districts have experimented with technologies marketed as weapons detection, gunshot detection, facial recognition, crowd analytics and threat identification systems. Companies such as ZeroEyes, Flock Safety, Avigilon and others have pitched AI-powered tools as ways to help under-resourced police and security departments respond faster to threats.
The appeal is obvious. Universities are under pressure to respond to mass shootings, sexual assaults, hate crimes, thefts, vandalism, protests, medical emergencies and unauthorized access. Parents want safe campuses. Administrators want faster investigations. Police departments want tools that reduce search time and increase situational awareness.
But critics argue that safety framing can obscure the long-term consequences of building permanent surveillance infrastructure into daily campus life. Once a camera system is purchased, installed and normalized, the debate often shifts from whether it should exist to which features should be activated next. That is where SDSU’s current assurances may not satisfy critics.
If facial recognition is not being used, will SDSU formally ban it? If behavioral tracking is not being used, will the university define that term and prohibit it? If license plate recognition is not active, will SDSU disclose whether the hardware or software can support it? If the system is only for motion alerts and maintenance, why was a more expansive AI-enabled platform purchased? If students are expected to accept cameras in residence halls, why not explicitly disclose AI capabilities in housing documents before they sign?
Those questions are not anti-safety. They are governance questions. A stronger SDSU policy would answer them directly.
At minimum, students and faculty should know where cameras are generally located, what kinds of analytics are enabled, who can access live and recorded footage, how long recordings are stored, whether footage can be shared with outside law enforcement, whether immigration authorities can access it, whether vendors can access or process the data, whether footage is used in student discipline, whether AI searches are logged, and whether the campus community receives annual public reports.
SDSU should also clarify whether any camera system can be used to monitor protests, identify participants in demonstrations, create face watchlists, search for people by clothing or physical appearance, track vehicles, count crowds, detect sounds, or produce automated alerts based on behavior.
The answer may be no. But for a public university, the answer should be written into policy rather than left to a spokesperson’s email.
The Daily Aztec investigation is important because it shows the continuing value of student journalism at a time when public institutions increasingly adopt technologies faster than communities can debate them. Without records requests and reporting by SDSU students, the public would know far less about the size, cost and capabilities of the campus camera network.
SDSU may argue that cameras make campus safer. That may be true in some situations. Surveillance footage can help solve crimes, locate missing people, reconstruct accidents and respond to emergencies. But public safety and privacy are not mutually exclusive. The question is whether the university is willing to impose real limits on itself before the technology is abused, rather than after controversy forces reform.
For now, SDSU has more than 1,300 AI-enabled cameras across campus and says the most invasive tools are not being used. Students, faculty and staff are being asked to trust that those limits will hold.
In San Diego, a city already shaped by smart streetlight fights, license plate reader battles, border surveillance and AI policing debates, that may no longer be enough.
Originally published on May 30, 2026.
SDSU’s vendor, Avigilon, a Motorola Solutions company, markets AI video analytics as a way to detect anomalies, automate investigations and help security personnel respond more quickly. Avigilon documentation also describes facial recognition lists and related tools that can allow authorized users to create profiles and search for faces across enabled cameras.
SDSU says it is not using facial recognition or behavioral profiling. That statement is important, but it does not eliminate the policy question. Once the infrastructure exists, future use can change through software settings, administrative decisions, policy revisions, budget pressure, law enforcement demands or leadership turnover. Privacy advocates often call this “mission creep,” the process by which a technology introduced for one limited purpose gradually expands into broader uses.
The Daily Aztec’s earlier reporting found that SDSU documents described desired capabilities beyond simple maintenance alerts, including AI-based anomaly detection, license plate readers, gunshot detectors and the ability to search video footage for physical characteristics. University police have since said those more sensitive capabilities are not being accessed.
That leaves SDSU in a familiar position for institutions adopting surveillance technology: asking the public to trust that powerful tools will remain unused because current administrators say they will not activate them.
San Diego has already been through this debate. The city became a national example of surveillance backlash after thousands of “smart streetlights” were installed beginning in 2016 under a program initially presented as a tool for traffic, parking and environmental data. Residents later learned the streetlights also contained cameras that police could access for investigations. During the 2020 Black Lives Matter protests, San Diego police used the streetlight camera system dozens of times while investigating alleged vandalism, looting and property destruction.
The backlash helped fuel the city’s Transparent and Responsible Use of Surveillance Technology Ordinance, known as the TRUST Ordinance, which requires city departments to disclose surveillance technologies, publish impact reports, submit use policies and obtain public approval. The ordinance also created the city’s Privacy Advisory Board.
Even with those safeguards, San Diego’s surveillance fights have not stopped. The city’s use of Flock Safety automated license plate readers and smart streetlight technology has continued to face opposition from privacy advocates, immigrant rights organizations and civil liberties groups. The Electronic Frontier Foundation and local TRUST coalition members have warned that license plate reader networks collect information on ordinary people’s movements, can be shared improperly, and can chill First Amendment activity.
San Diego’s citywide fight is not identical to SDSU’s campus camera system. SDSU is governed by CSU policy, not the city’s TRUST Ordinance. But the underlying issue is similar: powerful surveillance technologies are often installed before communities fully understand how they work, how data is retained, who can access it, whether third parties can obtain it, and what remedies exist if the technology is misused.
For students, the stakes are unusually personal. A public university is not just a workplace or a municipal street grid. It is a place where students live, study, socialize, protest, seek medical help, attend religious and cultural events, meet with counselors, join advocacy groups and move through formative years of adult life. A camera network spread through campus can capture not only safety incidents but also daily patterns of association and behavior.
That raises First Amendment concerns. Courts have long recognized that surveillance of political activity can chill speech and association, even when no one is arrested. On a college campus, where protests, student organizing and controversial speech are core parts of public life, students may reasonably wonder whether attending a demonstration, meeting with an activist group, participating in a labor action or joining a contentious campus debate could place them in searchable video records.
It also raises Fourth Amendment questions, though those are more complicated. In general, people have reduced privacy expectations in public spaces. Cameras in hallways, plazas and parking lots are not automatically unconstitutional. But AI changes the practical analysis. The constitutional issue is not only whether a person can be seen in public. It is whether the government can use technology to aggregate thousands of moments into a searchable record of a person’s movements, associations and habits. That is the same concern now being litigated and debated nationally around automated license plate readers, facial recognition, drone surveillance and other AI-assisted policing tools.
California law also complicates the picture. California is a two-party consent state for confidential audio communications, and CSU policy says video security cameras installed under the systemwide policy may not capture audio. If AI-enabled cameras include audio detection capabilities as a technical feature, SDSU would need to ensure those features are not used in a way that violates state law or CSU policy.
Student records law may also come into play. Under federal FERPA guidance, a school surveillance video can become an education record when it is directly related to a student and maintained by the institution, such as when footage is used in student discipline. CSU’s own policy acknowledges that video camera recordings involving specific students are generally considered law enforcement records unless the university uses them for student discipline or makes them part of an education record. That means the same footage could occupy different legal categories depending on how SDSU uses it.
The campus labor implications are also significant. CSU policy states that video security cameras shall not be used for employee supervision, timekeeping, attendance, evaluation or general monitoring of employee conduct. It allows footage to be used to corroborate, prove or disprove specific acts of misconduct under certain circumstances, but prohibits routine use as a workplace surveillance tool.
That is an important safeguard, but it depends on documentation, access controls and enforcement. A policy that says cameras will not be used for employee supervision is only as strong as the logs, audits and consequences that accompany it.
SDSU’s situation also comes during a broader national expansion of AI-enabled security on college campuses. Michigan State University has pursued a system designed to detect barrier breaches, track individuals, count crowd sizes and read license plates. Other campuses and school districts have experimented with technologies marketed as weapons detection, gunshot detection, facial recognition, crowd analytics and threat identification systems. Companies such as ZeroEyes, Flock Safety, Avigilon and others have pitched AI-powered tools as ways to help under-resourced police and security departments respond faster to threats.
The appeal is obvious. Universities are under pressure to respond to mass shootings, sexual assaults, hate crimes, thefts, vandalism, protests, medical emergencies and unauthorized access. Parents want safe campuses. Administrators want faster investigations. Police departments want tools that reduce search time and increase situational awareness.
But critics argue that safety framing can obscure the long-term consequences of building permanent surveillance infrastructure into daily campus life. Once a camera system is purchased, installed and normalized, the debate often shifts from whether it should exist to which features should be activated next. That is where SDSU’s current assurances may not satisfy critics.
If facial recognition is not being used, will SDSU formally ban it? If behavioral tracking is not being used, will the university define that term and prohibit it? If license plate recognition is not active, will SDSU disclose whether the hardware or software can support it? If the system is only for motion alerts and maintenance, why was a more expansive AI-enabled platform purchased? If students are expected to accept cameras in residence halls, why not explicitly disclose AI capabilities in housing documents before they sign?
Those questions are not anti-safety. They are governance questions. A stronger SDSU policy would answer them directly.
At minimum, students and faculty should know where cameras are generally located, what kinds of analytics are enabled, who can access live and recorded footage, how long recordings are stored, whether footage can be shared with outside law enforcement, whether immigration authorities can access it, whether vendors can access or process the data, whether footage is used in student discipline, whether AI searches are logged, and whether the campus community receives annual public reports.
SDSU should also clarify whether any camera system can be used to monitor protests, identify participants in demonstrations, create face watchlists, search for people by clothing or physical appearance, track vehicles, count crowds, detect sounds, or produce automated alerts based on behavior.
The answer may be no. But for a public university, the answer should be written into policy rather than left to a spokesperson’s email.
The Daily Aztec investigation is important because it shows the continuing value of student journalism at a time when public institutions increasingly adopt technologies faster than communities can debate them. Without records requests and reporting by SDSU students, the public would know far less about the size, cost and capabilities of the campus camera network.
SDSU may argue that cameras make campus safer. That may be true in some situations. Surveillance footage can help solve crimes, locate missing people, reconstruct accidents and respond to emergencies. But public safety and privacy are not mutually exclusive. The question is whether the university is willing to impose real limits on itself before the technology is abused, rather than after controversy forces reform.
For now, SDSU has more than 1,300 AI-enabled cameras across campus and says the most invasive tools are not being used. Students, faculty and staff are being asked to trust that those limits will hold.
In San Diego, a city already shaped by smart streetlight fights, license plate reader battles, border surveillance and AI policing debates, that may no longer be enough.
Originally published on May 30, 2026.
