VPD facing backlash over possible use of AI in post online

The Vancouver Police Department (VPD) is being called out after it posted an online photo that some say appears to have been altered by artificial intelligence (AI).

On Tuesday, the VPD posted a news release that said it had recently made nine arrests related to alleged drug trafficking. In the release, an image of the items seized was reportedly included.

This post is being questioned by some people online over what they say are inconsistencies in the images — distorted lettering and what appear to be fake bills, among others — with some noting the original posting on Facebook contained the tag “AI info.”

In a statement to CityNews, VPD Sgt. Adam Donaldson said they used software to edit out the names of the accused, but did not mention AI.

“The edited photo has been taken down and replaced with the original photo that has been cropped to exclude the names of the accused,” he said.

He confirms the following image was the one that was posted by mistake:

---

---

Altered images can erode public confidence in the police: expert

Sharon Bauer, founder of Bamboo Data Consulting, is a privacy consultant and AI governance strategist.

She says police often post images that have been tweaked to protect people’s privacy.

However, she says, they don’t usually use AI.

“That is just a different technology than they otherwise would be using,” Bauer said.

“When using AI, of course, there could be mistakes, there could be hallucinations, as we saw in the photo that was released.”

This is concerning, particularly because it erodes the public’s confidence in the police when things like this happen, she says.

“They are the organization that the public should trust and needs to trust,” she said.

“…the police didn’t have proper governance in place or policies in place to make sure that before disclosing any images containing evidence or anything other than that, they may not be using AI responsibly and there may not be someone viewing these images before they are released to the public,” she said.

“We’re looking at internal governance as opposed to actual laws.”

Bauer says there are no AI laws that require them to do anything when posting these altered images.

“That’s a major gap currently in our system, and what they really need to do is create their own rules, clear rules and policies about when images can be edited or what kinds of edits are allowed and how those edits are disclosed to the public,” she said.

“We’re looking at internal governance as opposed to actual laws.”

In other words, coming up with their own rules and policies would be a voluntary process, she says.

This isn’t true everywhere, however.

“In different provinces, there are different laws that just require, for example, transparency when AI is being used,” Bauer said.

“For example, in Ontario, governmental organizations will need to disclose when using AI, so there is an element of transparency. So in that regard, we do see law trying to catch up, but it certainly isn’t there in all provinces and in within all organizations.”

Many are left wondering what to do, she says.

“They’re trying to figure things out on their own, and getting some more guidance and structure around how to use AI responsibly and in a trustworthy way would be very helpful in getting some not just guidance, but very kind of prescriptive rules around that,” Bauer said.

“So it’s not a guessing game of can we do this? Can we not? Have we done enough or have we not?”

Meanwhile, the VPD’s post has since been replaced with one that shows the new image.

---

---