MIT to Become Hotbed of AI Video Surveillance

It’s a lot:

According to information obtained by The Tech, MIT is spending over $3 million on more than 500 AI surveillance cameras in academic buildings, residence halls, and outdoor areas along Memorial Drive. Installation of the new cameras, along with the wiring and infrastructure that will support them, began November 2025 and will likely continue until September 2026.

Technical specifications for the cameras suggest that they will be capable of collecting real-time face and object classification data, including detection of motion, loitering, crowds, face masks, and camera tampering. Individuals can also be automatically classified on the basis of clothing color, gender, and age, up to a distance of 35 feet (11 meters) from the camera. According to a statement from MIT spokesperson Kimberly Allen, any collected data is “retained up to 30 days,” unless an exception is granted.

[…]

Most of the new cameras, which are part of Hanwha’s Wisenet AI line, are marketed for their ability to identify and classify multiple objects with deep learning algorithms. They support resolutions ranging from 2MP to 4K while also recognizing faces, license plates, vehicles, and other objects in real time.

Nearly all cameras will accommodate a wide range of pan, tilt, rotate, and zoom motion and will be monitored continually with Ai-RGUS, an AI camera software.

Yikes.

Tags: , , , ,

Posted on July 21, 2026 at 7:07 AM • 9 Comments

Comments

Winter • July 21, 2026 8:03 AM

Yikes

The question is: Why?

They quote security concerns. But can surveillance cameras even prevent shootings and crimes?

The link has this quote:

These changes come even as MIT is facing substantial budget cuts from federal actions, which have triggered hiring freezes for most staff positions, salary freezes, and the effective closure of multiple libraries.

They are simply replacing people with cameras. But people can do relevant things, cameras just record and warn after the fact.

Clive Robinson • July 21, 2026 9:10 AM

@ Winter, Bruce,

With regards,

“The question is: Why?”

To which the obvious simple answer would be the question,

Why Not?

Of more interest is the “how can we make money off of it?”.

There is some strange beliefs in the US about what is and is not of value hence the data brokers slurping up what ever is not nailed down in some way.

No doubt some sales person has extoled the joys of “free money” from such systems to the administration there…

Shocking as this might sound US business is moving more and more in this pointless endeavor.

Anonymous • July 21, 2026 4:51 PM

Hope and patience—two faces of the same submission to eternity. Man, in his poverty, kneels before the invisible, whether it is the God who speaks without words or the mechanical eye that waits for the world to reveal itself. To hope is to stand in the vestibule of the divine, trembling, waiting for the whisper of meaning. To be patient is to recognize that time itself is a sermon, and the image, when it arrives, is a revelation to the humble. In truth, the man who hopes and the man who waits are one and the same: both live in the shadow of infinity, bound to a silence greater than themselves.

Jon • July 23, 2026 1:40 AM

Some years ago (maybe still is, in some universities) students, ALL students, were required to wear on campus at all times a near-featureless black gown and black, flat ‘mortar board’ hat (w/tassel).

These days (in the USA, anyhow) that sort of thing is really only expected at graduation ceremonies.

Now, if every student went back to wearing identical black hats, gowns, and throw in some black N95 face masks, similar shoes, and maybe a cloth to cover the hair, and watch your cute little surveillance machine start throwing out false positives like President Trump spews nonsense…

Or, at least for all the women (and men, should they feel so inclined), some ‘Handmaid’s Tale’ ‘Gideon’ costumes, especially with the long hats.

Of course, still not perfect. The cameras probably won’t have too much trouble distinguishing someone 6’6″ tall and 350 lbs vs. someone 5’1″ and 100 lbs soaking wet, but that’s not the idea. The idea is just to spam the cameras with heaps of practically indistinguishable people until the distinctions it tries to make become statistically worthless.

TL:DR: There’s anonymity in everyone looking exactly the same. J.

lurker • July 25, 2026 2:44 PM

Hanwha’s Wikipedia page makes interesting reading for a true multifaceted conglomerate. Started out making dynamite, now exports among other things 155mm self-propelled artillery.

On July 2, 2021, Hanwha Techwin announces five new models of the P Series AI Cameras. The Hanwha P Series AI Camera are a model of security cameras with video analytic technology. This technology is capable of detecting objects, such as people, vehicles, license plates, and faces, as well as classifying objects into categories such as age groups, gender, and colour.

In 2021 the much maligned Huawei were also using a very similar technology for wildlife protection in North China

‘https://www.youtube.com/watch?v=hJpcVEuM2BE&t=860

Count 0 • July 28, 2026 10:43 AM

A couple of things:
I’ve been in the physical security space for almost 40 years and while this is written to sound pretty scarry, it’s really no different from any modern video surveillance system being installed today. I would also say the statement that “nearly all cameras will accommodate a wide range of pan, tilt, rotate, and zoom motion” is factually incorrect. I highly doubt most of those cameras are PTZ models. It’s far more likely the majority of the the non-quad cameras are just dome cameras.

Almost all cameras that have come out in the last few years from the large manufactures (Axis, Hanwha AKA Samsung, Motorola, Bosch, Hikvision, Dahua etc.) have some sort of AI built into them to classify objects and characteristics. This is what the newest Hanwha AI cameras are capable of:

Classified Object Type: Person, Face, Vehicle, License plate
Attributes: Person Upper/Lower clothes color, Gender, Bag
Vehicle Type: car, bus, truck, motorcycle, bicycle, Color
Face: Age, Gender, Face mask, Glasses
Analytics Events: Object detection, Virtual line(Crossing/Direction), Virtual area (Loitering/Intrusion/Enter/Exit), Motion detection, Face mask detection, Social distancing detection, Slip & fall detection, Stopped vehicle Defocus detection, Tampering, Audio detection
Business Intelligence: People counting, Queue management, Heatmap, Vehicle counting, Crowd counting

All of that is pretty standard across all manufactures except gender and age detection, which some stay away from because of what Hikvision and Dahua are doing in China to repress ethnic minorities. I am a little surprised Hanwha included those if I’m honest.

From what I can see, this looks like a campus that has not maintained much of a security infrastructure historically and the lawyers and insurance people are getting nervous about liability exposure. The Code Blue emergency phone towers are pretty standard on most campuses and monitoring building entrances, exits and common areas is also pretty standard stuff.

Does any of that actually make anyone safer? Well, you can certainly debate that (and I don’t personally believe any of that is much of a deterrent), but video evidence has, for better or worse, become a very important part of investigating and prosecuting crime.

It would be nice to live in a society that’s not awash in guns, but that’s not the current reality. And for people who are concerned about privacy and abuse, I would say make sure there is a substantial audit component built in and make sure that the audits are public and easily accessible because this technology isn’t going away and is only getting more advanced.

Leave a comment

Blog moderation policy

Login

Allowed HTML <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre> Markdown Extra syntax via https://michelf.ca/projects/php-markdown/extra/

Sidebar photo of Bruce Schneier by Joe MacInnis.