Because it's an active construction site, only 1/4 planned recreation areas was built, so it had to be shared by 1200+ detainees. They each got 40 min recreation a few times a week. "This lack of recreation time is adding to the detainees' visible frustrations and unruliness" the inspectors wrote.

This month, ICE inspectors toured the giant new Fort Bliss tent camp for migrants, built in under two months.

They found 60 violations of federal detention standards, many of which directly impacted the safety of workers and detainees, we reported today in @washingtonpost.com

🧵-->

Explore ICE's planned facilities in your region or download the entire list of 125 new and expanded facilities using the handy table inside our story: wapo.st/45hE9fq

Other notable facilities slated to open:

-Puerto Rico's only large-scale ICE detention center
-Minnesota's only private prison
-California City facility will become largest ICE center in CA
-Irwin, a Georgia detention center closed after women alleged medical abuse
-Military bases in NJ & IN

Trump's ICE believes bigger is better. Number of detention centers over 1,000 beds:

Now: 29
End of year: 49

Photo: Ft. Bliss soft-sided detention site in Texas, expected to hold up to 5,000 migrants. (REUTERS/Paul Ratje)

Trump plans to open at least two new family detention facilities, more than doubling the record for total family capacity set under the Obama and first Trump administrations.

ICE also plans to expand in states with few existing detention beds, including Oklahoma, Indiana, Minnesota, North Carolina and Tennessee.

Number of states that can hold over 1,000 ICE detainees:

Now: 14
End of year: 19

Four states now account for the majority of ICE’s detention space: Texas, Louisiana, California and Georgia. That will continue, with new facilities in Texas alone expected to double the state’s capacity to an incredible 38,000 beds by year-end.

Congress allocated $45 billion to spend locking up immigrants over the next four years — more than the government spent on ICE detention during the Obama, Biden and first Trump administrations combined. Story: www.washingtonpost.com/business/202...

The policing patterns appear widespread.

Among 23 police departments with detailed arrest records on facial recognition cases, 15 made at least one arrest of an AI suspect with no independent evidence.

Many appear to have violated their own policies which require corroboration of AI results.

We know of 8 wrongful arrests (several uncovered by good reporting of @kashhill.bsky.social) but the real number is undoubtedly higher bc police mostly use AI software in secret.

+ these people all say they got lucky: they had a mole or tattoo that proved innocence, they had money for lawyer, etc

Even when AI works, it can lead to embarrassing mistakes.

Jason Vernau was held for 3 days after police mistook him for a man who cashed a fraudulent check. Facial recognition correctly ID'd Vernau from this image, but police failed to realize the image was just a bank customer, not the criminal.

Police in Woodbridge, NJ were so intent on building a case against Nijeer Parks, a robbery suspect they found through facial recognition, that they cast aside DNA and fingerprint evidence pointing to another potential suspect.

Parks eventually went free but the other guy has never been charged.

That confidence in computer software seems to lead police to skip basic police work, such as checking alibis, taking fingerprints and checking tattoos. Porcha Williams was arrested for carjacking despite being 8 mos pregnant at the time and no indication the crime was committed by a pregnant woman

Police say they only treat facial recognition as an "investigative lead" but records show they often view AI results as proof of guilt.

Here's one arrest report that referred to an uncorroborated facial recognition result as a "100% match"