This is the current issue of The Download, our daily newsletter delivering a regular update on the happenings in the tech world.
An AI model developed using prison phone calls is now detecting potential crimes in those communications
A telecommunications firm in the US has trained an AI model on years’ worth of incarcerated individuals’ phone and video interactions and is currently testing this model to scrutinize their calls, texts, and emails with the aim of forecasting and averting criminal activities.
Kevin Elder, president of Securus Technologies, informed MIT Technology Review that the company initiated the development of its AI technologies in 2023, leveraging its extensive archive of recorded conversations to train AI models capable of identifying illegal behaviors. For instance, one model was developed using a dataset of seven years of calls made by inmates within the Texas correctional system, and efforts are ongoing for models applicable in additional states and counties.
Nonetheless, advocates for prisoner rights contend that this new AI framework facilitates a type of intrusive monitoring, and judicial authorities have set few constraints on this authority. Read the complete article.
—James O’Donnell
Nominations are now open for our global 2026 Innovators Under 35 challenge
We have some thrilling news: Nominations are now being accepted for MIT Technology Review’s 2026 Innovators Under 35 challenge. This annual list celebrates 35 of the world’s most outstanding young scientists and creators, a tradition upheld by our newsroom for over two decades.
Submitting a nomination for yourself or someone you know is free and takes just a few moments. Here’s how to submit your nomination.
The essential reads
I’ve scoured the web to bring you today’s most entertaining/significant/alarming/captivating narratives about technology.
1 New York is tightening regulations on personalized pricing algorithms
A new law requires retailers to disclose whether their pricing is influenced by consumer data. (NYT $)
+ The US National Retail Federation attempted to prevent its passage. (TechCrunch)
2 The White House has introduced a media bias monitoring tool
Featuring a “media offender of the week” section along with a Hall of Shame. (WP $)
+ The Washington Post is currently identified as the site’s top offender. (The Guardian)
+ Donald Trump has criticized several journalists in recent weeks. (The Hill)
3 American startups are relying on open-source Chinese AI models
They’re affordable and adaptable—what’s not to appreciate? (NBC News)
+ Americans also favor China’s inexpensive products, regardless of tariffs. (WP $)
+ The State of AI: Is China poised to win the competition? (MIT Technology Review)
4 How police body camera footage became viral YouTube material
Newly arrested individuals dread the thought of appearing on popular channels. (Vox)
+ AI was intended to enhance police body cameras. What went wrong? (MIT Technology Review)
5 Construction workers are benefiting from the data center surge
May as well enjoy it while it lasts. (WSJ $)
+ The data center surge in the desert. (MIT Technology Review)
6 China remains skeptical about crypto
Despite bitcoin mining quietly making a (banned) resurgence. (Reuters)
+ The nation’s central bank has little regard for stablecoins. (CoinDesk)
7 A startup is treating its AI partners as if they were characters in a story
Could such a strategy lead to enhanced AI companions? (Fast Company $)
+ Gemini has emerged as the most empathetic model, it seems. (Semafor)
+ The impending restrictions on AI companionship. (MIT Technology Review)
8 Ozempic is outdated 💉
New weight-loss medications are customized for individual patients. (The Atlantic $)
+ What is still unknown about weight-loss medications. (MIT Technology Review)
9 AI is transforming consultancy work
For the third consecutive year, major firms are maintaining salary freezes for junior employees. (FT $)
10 Behind the curtain of Disney’s AI animation accelerator
What took five months to produce has now been reduced to less than five weeks. (CNET)
+ Director extraordinaire James Cameron appears to have shifted his perspective on AI. (TechCrunch)
+ Why are individuals perusing oddly-formatted TV clips? (WP $)
Quote of the day
“[I hope AI] reaches a stage where it becomes akin to mental junk food and we feel unwell without understanding the cause.”
—Actress Jenna Ortega expresses her aspirations for AI’s forthcoming role in cinema, Variety reports.
One additional note
The weeds are prevailing
Since the 1980s, an increasing number of plants have adapted to become resistant to the biochemical methods that herbicides use to eliminate them. This resistance poses a risk of diminishing crop yields—uncontrolled weeds can lead to a reduction of 50% or more in yields, with extreme instances potentially wiping out entire fields.
In the worst-case scenario, it could even drive farmers to bankruptcy. It mirrors the agricultural crisis of antibiotic resistance, continuously worsening. Weeds have acquired resistance to 168 distinct herbicides and 21 out of the 31 recognized “modes of action,” referring to the specific biochemical target or pathway that a chemical is designed to affect.
Agriculture must adopt a variety of weed management strategies. However, implementing this is much simpler in theory than in practice. Read the complete article.
