Philadelphia-area school district leader reflects on new AI safety system added this year
Some schools in Upper Darby have a new security system powered by artificial intelligence
Some schools in Upper Darby have a new security system powered by artificial intelligence
As the 2024–2025 school year wraps up, the superintendent of Delaware County's largest school system is reflecting on the impact of new technology designed to keep students safe.
Upper Darby School District Superintendent Daniel McGarry said since its installation on Dec.11, 2024, Evolv has detected a total of seven knives or box cutters, two brass knuckles and seven pepper spray devices across the three schools.
"We don't have any evidence that these weapons were coming on campus as a use of force," McGarry said. "It was usually an awkwardness like, 'I had this because I was cutting food.'"
Evolv, which utilizes artificial intelligence to scan for potential weapons, was installed at Upper Darby High School, Beverly Hills Middle School and Drexel Hill Middle School.
"There was fear and some trepidation that it would take hours to get kids into the school, that we would over-identify and improperly search kids," McGarry said. "Thanks to the work of our kids and our school community, none of those things actually happened."
When staff found weapons, they confiscated them and turned them over to police. Police conducted a background check on families and, in some cases, a threat assessment. Students were disciplined for having a weapon on school property, but they were not determined to be a threat in any of the cases.
Installing a weapons-detection system had been in the works for years, but the idea gained renewed interest after two firearms-related incidents in the district at the beginning of the school year.
McGarry said a student brought a gun to Upper Darby High School on Sept. 3, 2024, and then 10 days later, another student brought in an AR-15-style magazine.
McGarry previously told CBS News Philadelphia the weapons detection system is "not a metal detector." The superintendent said it's the same technology used at Lincoln Financial Field.
"Since we had the gun in the beginning of the school year, we have not found any guns on any individuals on campus at any of our secondary schools," McGarry said.
Certain harmless objects like Chromebooks are known to set off the scanners.
"We had almost 800,000 visitors from Dec. 11 to the last time we gave this update, which was this past month," Superintendent McGarry said. "Of those 800,000 visitors, there were 70,000 alerts. The vast majority of those alerts were for umbrellas, computers left in backpacks and eyeglass kits."
McGarry said he is confident that as students continue to get used to the technology, they will be more forthcoming about removing their laptops and eyeglass cases from their backpacks.
"To put anything in the middle of the school year, just weeks before you go on winter break, and to have the success we have, speaks to the type of people who work at this organization," McGarry said.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
CBS News
an hour ago
- CBS News
91-year-old Pittsburgh-area woman dies after own car rolls over her
A 91-year-old woman in Cranberry Township died on Friday after her vehicle apparently rolled over her, police said. Cranberry Twp Police Lt. Chuck Mascellino said crews were called to the parking lot of the Post Office on Executive Drive around 10 a.m. for an incident. The officer said police believe the 91-year-old driver did not put her car into park before getting out, and it may have rolled over her. The woman was taken to a local hospital for treatment. She later died at the hospital. Mascellino added that police believe the car hit her, but there could have also been a medical incident. The official cause is under investigation. No other information was released on Friday.
Yahoo
an hour ago
- Yahoo
Seller Experience Is ThredUp's Next Frontier for AI
ThredUp's attributing some of its success to artificial intelligence. That it boasts about its myriad tools is far from a surprise—James Reinhart, the company's CEO and co-founder has taken an all-in mentality on the technology for about 18 months now. To date, most of ThredUp's innovations have focused on its buyers, rather than its sellers. More from Sourcing Journal Byte-Sized AI: Debenhams Launches Employee Upskilling Program; Aetrex's New AI-Enabled Foot Scanner NY AG Asks Court to Dismiss NRF's Algorithmic Pricing Lawsuit Six in 10 Transportation Management Leaders Believe Agentic AI Will Act on Their Behalf by 2030 According to Reinhart, that strategy seems to be paying off when it comes to customer acquisition and conversion. In its earnings call this week, the company indicated it had increased new buyer acquisition by 74 percent year on year, while also increasing conversion by 18 percent year on year. While Reinhart said on the earnings call that the company doesn't have an exact breakdown of which tools drive the highest conversion rates, he told Sourcing Journal that he believes the company's Shop Similar feature, which provides similar recommendations powered by AI, has been 'a big driver' for the conversion increase. He called the company's Image Search tool and other AI-powered features 'incrementally positive.' '[Shop Similar] nails the experience. You're looking for a shirt, and you find a shirt that you like and you're like, 'Ah, I want it in a different color.' You used to have to start over, but now it's like, 'Show me it in different colors,'' he said. 'It's really created this nice way for people to stay in the shopping journey, versus having to always start over.' Reinhart said Style Chat, ThredUp's conversational AI interface, which helps users put together occasion-based outfits and responds to longer, natural-language queries, and its social commerce tool, which ingests users' Pinterest feeds to help provide recommendations, are both 'advanced feature[s].' He noted that users who engage with the tools regularly convert at 'much higher rates' but said that many users have yet to discover the power of those tools. 'It's something that people, I think, are just starting to use in their lives, the way they might be using ChatGPT,' he said. And though ThredUp has a conversational interface of its own, Reinhart said he's keenly aware that the company's inventory needs to be publicly accessible to third-party AI models, like OpenAI's ChatGPT, which offers users a shopping function. He said the team is investing in resources to make the backend of ThredUp's site fully crawlable for other companies to keep pace with consumers' preferred ways of shopping.'I think [LLMs are] as big, or bigger than, the impact of SEO. In the SEO world, if you appear at the top of the rankings, you got all the traffic…I think the same is true [here]. I think if you don't appear in ChatGPT or Anthropic—whatever the agent is, if you're not appearing in that list, I think you're going to get left behind.' Much of the company's AI movement in recent months has been targeted at raking in new consumers; Reinhart said that, while that's always a goal for the business, longer-term ThredUp buyers also seem to have less interest in integrating AI into their shopping experiences. For instance, he said, this year, the company launched on-model photography by leveraging AI. In an A/B test, new customers flocked to items with listings showcasing on-model photos, but existing customers didn't seem moved by the test.'Our existing customers were like, 'I don't need that,' because they're all in on ThredUp. There's not a lot we can do on the innovation side where they're going to be blown away, because they're already committed,' he said. For Reinhart, that means it's time to leverage technology to go back to the basics. He said he knows repeat customers simply want to find quality pieces at an affordable price; many have already mastered the art of sifting through the millions of listings ThredUp offers on any given day. 'The biggest way to delight our heaviest users is to have ever better assortment, because they already know how to shop it,' Reinhart said. 'They just need more stuff to buy from us that meets their needs, and so that's why I'm balancing seller innovation as well as new buyer innovation.' To do that, he contends, it's time to shift the organization's focus a bit toward developing AI that can better serve sellers, rather than fixating on buyers. 'The seller experience on ThredUp had, for the prior few years, taken a backseat to focus on the buyer journey—literally, every single AI tool is focused on buyers. So, I think now, we're thinking, 'Okay, well, how can we use AI to improve the seller experience? What types of products can work better for sellers?'' He said the company plans to 'tighten up' how sellers view items, how they handle cleanout kits and more. Reinhart said the process of redefining the seller experience has already begun with ThredUp Premium, which charges a higher fee for listing items but typically yields higher return and longer selling window times for the seller. On the earnings call, he said those ordering premium kits increased 44 percent quarter on quarter. But seller innovation and buyer innovation could soon dovetail more seriously; Reinhart hinted at a tool that would allow human creators and influencers to leverage AI to promote specific products on the ThredUp site. While that could help buyers select items, it could also help sellers move products quicker. 'You're going to continue to see us invest in [what] I would describe as using AI and human creators together to tell powerful merchandising stories. If I were to, for example, give you some of the internal tools at ThredUp to put together outfits or curations…you could build something that people would want to shop,' he told Sourcing Journal. 'What would it look like to enable creators and influencers to tell stories, to curate their own list of great content? We're going to work a bunch on that.' ThredUp's ambitious AI goals require some manpower to build, test, iterate on and push to market. In 2024, the company laid off about 25 percent of its staff, and Reinhart later disclosed that a penchant for investing in AI was part of the reason for those cuts. Now, he said, the company has added heads in a targeted way. 'We have been hiring, really with a focus on people who have skill sets that can utilize either the emerging tools in AI or they're naturally curious about how it works,' he said, noting that the company has 'a few more key roles to hire for' before pausing its hiring. 'Our pace of hiring will probably slow a little bit into [2026]. I think we have staffed up to a level where we're launching new products and innovating at a pretty healthy rate, and I'd like to see us breathe a little with the resources that we have,' he added.
Associated Press
2 hours ago
- Associated Press
'Devil in Ozarks' planned prison escape for months, cited lax security in kitchen, report says
LITTLE ROCK, Ark. (AP) — A former police chief known as the 'Devil in the Ozarks' spent months planning his escape from an Arkansas prison, and said lax security in the kitchen where he worked allowed the convicted murderer to gather the supplies he needed, an internal review by prison officials released Friday said. The Department of Corrections' critical incident review of Grant Hardin's May 25 escape from the Calico Rock prison provides the most detailed description so far of his planning and the issues that allowed him to walk out of the facility. Hardin was captured 1.5 miles (2.4 kilometers) northwest of the Calico Rock prison on June 6. Authorities said he escaped by donning an outfit he designed to look like a law enforcement uniform. Hardin, who worked in the prison's kitchen, said he spent six months planning his escape and used black Sharpie markers and laundry he found lying around the kitchen to create the fake uniform, according to the report. Hardin fashioned a fake badge using the lid of a can. 'Hardin stated he would hide the clothes and other items he was going to need in the bottom of a trash can in the kitchen due to no one ever shaking it down,' the report says. Two prison employees have been fired for procedure violations that led to Hardin's escape. They include a kitchen employee who allowed Hardin on a back dock unsupervised and a tower guard who unlocked the back gate that Hardin walked through without confirming his identity. Several other employees have been suspended and one demoted, lawmakers were told this week. The kitchen's staff was 'very lax on security,' Hardin told investigators, allowing him to gather what he needed for his escape. Hardin said he didn't have any help from staff or other inmates. Hardin had constructed a ladder from wooden pallets in case he needed to scale the prison fence but didn't need it. '(Hardin) stated when he walked up to the gate, he just directed the officer to 'open the gate,' and he did,' the report says. After he escaped from the prison, Hardin survived on food he had smuggled out of the prison along with distilled water from his CPAP machine. Hardin also drank creek water and ate berries, bird eggs and ants. 'He said his plan was to hide in the woods for six months if need be and begin moving west out of the area,' the report says. Hardin, a former police chief in the small town of Gateway, near the Arkansas-Missouri border, is serving lengthy sentences for murder and rape. He was the subject of the TV documentary 'Devil in the Ozarks.' The report is one of two reviews into Hardin's escape, which is also being investigated by the Arkansas State Police. A legislative subcommittee has also been holding hearings about the escape. Republican Rep. Howard Beaty, who co-chairs the Legislative Council's Charitable, Penal and Correctional Institutions Subcommittee, said the panel hoped to discuss both reports with officials at a hearing next month. Republican Sen. Ben Gilmore, who sits on the panel, said he didn't think the department's review took a thorough enough look at the systemic issues that enabled Hardin's escape. 'They have focused on the final failure instead of all of the things that led up to it,' he said. The report also cites confusion among corrections officials in the early stages of Hardin's escape about which law enforcement agencies had been notified, the report says. 'It is obvious there was a lot of confusion during the beginning stages of opening the command center and of notifications being made,' the report says. Hardin had been misclassified and shouldn't have been held at the primarily medium-security prison, according to the review. After he was captured, Hardin was moved to a maximum-security prison. He has pleaded not guilty to escape charges, and his trial is set for November. Hardin's custody classification hadn't been reviewed since October 2019, the report says. The Department of Corrections' review says officials had taken several steps since Hardin's escape, including removing the electric locks from the gates to prevent someone from walking out without an officer present. The report also calls for additional cameras after finding a blind spot on the dock Hardin used, and for any 'shakedown' searches for contraband to include mechanical rooms and side rooms.
