datum

After more than two and a half years of research, planning and construction, Dataland, the world’s first museum of AI arts, will open June 20.

Co-founded by new media artists Refik Anadol and Efsun Erkılıç, the museum anchors the $1-billion Frank Gehry-designed Grand LA complex across the street from Walt Disney Concert Hall in downtown Los Angeles. Its first exhibition, “Machine Dreams: Rainforest,” created by Refik Anadol Studio, was inspired by a trip to the Amazon and uses vast data sets to immerse visitors in a machine-generated sensory experience of the natural world.

The architecture of the space, which Anadol calls “a living museum,” is used to reflect distant rainforest ecosystems, including changing temperature, light, smell and visuals. Anadol refers to these large-scale, shimmering tableaus as “digital sculptures.”

“This is such an important technology, and represents such an important transformation of humanity,” Anadol said in an interview. “And we found it so meaningful and purposeful to be sure that there is a place to talk about it, to create with it.”

The 35,000-square-foot privately funded museum devotes 25,000 square feet to public space, with the remaining 10,000 square feet holding the in-house technology that makes the space run. Dataland contains five immersive galleries and a 30-foot ceiling. An escalator by the entrance will transport guests to the experiences below. The museum declined to say how much Dataland, designed by architecture firm Gensler, cost to build.

An isometric architectural rendering of Dataland. The 25,000-square-foot AI arts museum also contains an additional 10,000 square feet of non-public space that holds its operational technology.

(Refik Anadol Studio for Dataland)

Dataland will collect and preserve artificial intelligence art and is powered by an open-access AI model created by Anadol’s studio called the Large Nature Model. The model, which does not source without permission, culls mountains of data about the natural world from partners including the Smithsonian, London’s Natural History Museum and the Cornell Lab of Ornithology. This data, including up to half a billion images of nature, will form the basis for the creation of a variety of AI artworks, including “Machine Dreams.”

“AI art is a part of digital art, meaning a lineage that uses software, data and computers to create a form of art,” Anadol explained. “I know that many artists don’t want to disclose their technologies, but for me, AI means possibilities. And possibilities come with responsibilities. We have to disclose exactly where our data comes from.”

Sustainability is another responsibility that Anadol takes seriously. For more than a decade, Anadol has devoted much thought to the massive carbon footprint associated with AI models. The Large Nature Model is hosted on Google Cloud servers in Oregon that use 87% carbon-free, renewable energy. Anadol says the energy used to support an individual visit to the museum is equivalent to what it takes to charge a single smartphone.

Anadol believes AI can form a powerful bridge to nature — serving as a means to access and preserve it — and that the swiftly evolving technology can be harnessed to illuminate essential truths about humanity’s relationship to an interconnected planet. During a time of great anxiety about the power of AI to disrupt lives and livelihoods, Anadol maintains it can be a revolutionary tool in service of a never-before-seen form of art.

“The works generate an emergent, living reality, a machine’s dream shaped by continuous streams of environmental and biological data. Within this evolving system, moments of recognition and interpretation emerge across different forms of knowledge,” a news release about the museum explains. “At the same time, the exhibition registers loss as part of this expanded field of perception, most notably in the Infinity Room, where visitors encounter the 1987 recording of the last known Kauaʻi ʻŌʻō, a now-extinct bird whose unanswered call becomes part of the work.”

“It’s very exciting to say that AI art is not image only,” Anadol said. “It’s a very multisensory, multimedium experience — meaning sound, image, video, text, smell, taste and touch. They are all together in conversation.”

WASHINGTON  — A man carrying a gun and a cellphone entered a federal credit union in a small town in central Virginia in May 2019 and demanded cash.

He left with $195,000 in a bag and no clue to his identity. But his smartphone was keeping track of him.

What happened next could yield a landmark ruling from the Supreme Court on the 4th Amendment and its restrictions against “unreasonable searches.”

Typically, police use tips or leads to find suspects, then seek a search warrant from a judge to enter a house or other private area to seize the evidence that can prove a crime.

Civil libertarians say the new “digital dragnets” work in reverse.

“It’s grab the data and search first. Suspicion later. That’s opposite of how our system has worked, and it’s really dangerous,” said Jake Laperruque, an attorney for the Center for Democracy & Technology.

But these new data scans can be effective in finding criminals.

Lacking leads in the Virginia bank robbery, a police detective turned to what one judge in the case called a “groundbreaking investigative tool … enabling the relentless collection of eerily precise location data.”

Cellphones can be tracked through towers, and Google stored this location history data for hundreds of millions of users. The detective sent Google a demand for information known as a “geofence warrant,” referring to a virtual fence around a particular geographic area at a specific time.

The officer sought phones that were within 150 yards of the bank during the hour of the robbery. He used that data to locate Okello Chatrie, then obtained a search warrant of his home where the cash and the holdup notes were found.

Chatrie entered a conditional guilty plea, but the Supreme Court will hear his appeal on April 27.

The justices agreed to decide whether geofence warrants violate the 4th Amendment.

The outcome may go beyond location tracking. At issue more broadly is the legal status of the vast amount of privately stored data that can be easily scanned.

This may include words or phrases found in Google searches or in emails. For example, investigators may want to know who searched for a particular address in the weeks before an arson or a murder took place there or who searched for information on making a particular type of bomb.

Judges are deeply divided on how this fits with the 4th Amendment.

Two years ago, the conservative U.S. Court of Appeals for the 5th Circuit in New Orleans ruled “geofence warrants are general warrants categorically prohibited by the 4th Amendment.”

Chief Justice John Roberts sided with the court’s liberals in a 4th Amendment privacy case in 2018.

(Alex Wong / Getty Images)

Historians of the 4th Amendment say the constitutional ban on “unreasonable searches and seizures” arose from the anger in the American colonies over British officers using general warrants to search homes and stores even when they had no reason to suspect any particular person of wrongdoing.

The National Assn. of Criminal Defense Lawyers relies on that contention in opposing geofence warrants.

Its lawyers argued the government obtained Chatrie’s “private location information … with an unconstitutional general warrant that compelled Google to conduct a fishing expedition through millions of Google accounts, without any basis for believing that any one of them would contain incriminating evidence.”

Meanwhile, the more liberal 4th Circuit in Virginia divided 7-7 to reject Chatrie’s appeal. Several judges explained the law was not clear, and the police officer had done nothing wrong.

“There was no search here,” Judge J. Harvie Wilkinson wrote in a concurring opinion that defended the use of this tracking data.

He pointed to Supreme Court rulings in the 1970s declaring that check records held by a bank or dialing records held by a phone company were not private and could be searched by investigators without a warrant.

Chatrie had agreed to having his location records held by Google. If financial records for several months are not private, the judge wrote, “surely this request for a two-hour snapshot of one’s public movements” is not private either.

Google changed its policy in 2023 and no longer stores location history data for all of its users. But cellphone carriers continue to receive warrants that seek tracking data.

Wilkinson, a prominent conservative from the Reagan era, also argued it would be a mistake for the courts to “frustrate law enforcement’s ability to keep pace with tech-savvy criminals” or cause “more cold cases to go unsolved. Think of a murder where the culprit leaves behind his encrypted phone and nothing else. No fingerprints, no witnesses, no murder weapon. But because the killer allowed Google to track his location, a geofence warrant can crack the case,” he wrote.

Judges in Los Angeles upheld the use of a geofence warrant to find and convict two men for a robbery and murder in a bank parking lot in Paramount.

The victim, Adbadalla Thabet, collected cash from gas stations in Downey, Bellflower, Compton and Lynwood early in the morning before driving to the bank.

After he was robbed and shot, a Los Angeles County sheriff’s detective found video surveillance that showed he had been followed by two cars whose license plates could not be seen.

The detective then sought a geofence warrant from a Superior Court judge that asked Google for location data for six designated spots on the morning of the murder.

That led to the identification of Daniel Meza and Walter Meneses, who pleaded guilty to the crimes. A California Court of Appeal rejected their 4th Amendment claim in 2023, even though the judges said they had legal doubts about the “novelty of the particular surveillance technique at issue.”

The Supreme Court has also been split on how to apply the 4th Amendment to new types of surveillance.

By a 5-4 vote, the court in 2018 ruled the FBI should have obtained a search warrant before it required a cellphone company to turn over 127 days of records for Timothy Carpenter, a suspect in a series of store robberies in Michigan.

The data confirmed Carpenter was nearby when four of the stores were robbed.

Chief Justice John G. Roberts, joined by four liberal justices, said this lengthy surveillance violated privacy rights protected by the 4th Amendment.

The “seismic shifts in technology” could permit total surveillance of the public, Roberts wrote, and “we decline to grant the state unrestricted access” to these databases.

But he described the Carpenter decision as “narrow” because it turned on the many weeks of surveillance data.

In dissent, four conservatives questioned how tracking someone’s driving violates their privacy. Surveillance cameras and license plate readers are commonly used by investigators and have rarely been challenged.

Solicitor Gen. D. John Sauer relies on that argument in his defense of Chatrie’s conviction. “An individual has no reasonable expectation of privacy in movements that anyone could see,” he wrote.

The justices will issue a decision by the end of June.

Prior to the Supreme Court’s 2015 Obergefell decision, opponents raised alarms about the severe and immediate harms that would surely occur if marriages between same-sex couples were recognized nationally. Afterward, when those harms failed to materialize, those voices grew quieter, but some have been returning with renewed vigor, in hopes that the current Supreme Court, after overturning Roe vs. Wade, may be willing to overturn the Obergefell decision as well — though the justices declined to do so in November.

To build public support for rolling back marriage rights, new campaigns have been repeating the claims that legal recognition of same-sex marriages may harm children or even the stability of different-sex marriages. These are some of the same concerns that were raised in the years prior to the Obergefell decision. They were groundless then, and, more than 10 years later, the data confirm these fears to be unfounded.

In 2024, for the 20th anniversary of the first legal marriages of same-sex couples (in Massachusetts), my lab at UCLA joined with a team of researchers at Rand Corp. to review what social scientists learned over those two decades about the consequences of legalizing same-sex marriage.

We addressed this question in two ways. First, we searched through the research literature to find every published study that had examined the consequences of legalizing same-sex marriage. Prior to 2015, states legalized and prohibited same-sex marriage at different times, and social scientists tracked a wide range of outcomes, including the well-being of children, national trends in marriage and divorce, and the physical and mental health of same-sex couples. Opponents of legalizing same-sex marriage predicted, in the strongest terms, that people would suffer after same-sex couples were granted the right to marry.

After 20 years of legalized marriage for same-sex couples, 96 independent studies confirm there is no evidence for the harms critics predicted. Our review identified not a single study that observed significant negative consequences of legalizing same-sex marriage. Instead, the research literature identified many significant positive consequences.

For same-sex couples, legal recognition of their marriages was followed by more stable relationships, increased mental and physical health, greater financial stability, and stronger connections to family. For the children of those couples, our review found no documented negative outcomes, but legal recognition of their parents’ marriages did result in more children obtaining access to health insurance. And what about the rest of the country? States that recognized same-sex marriages prior to Obergefell experienced economic gains and considerable savings in healthcare costs relative to states that did not.

One of the most striking predictions of the opponents of same-sex marriage was that recognizing marriage among same-sex couples would weaken commitment to the institution of marriage among different-sex couples. That did not happen either.

To address this question, our report conducted new analyses, drawing on census data and other sources to determine whether state-level rates of marriage, cohabitation and divorce changed in the states that recognized same-sex marriage, compared with states that did not. No matter how we conducted the analyses, we could find no effects of recognizing same-sex marriage on any of these outcomes. It makes sense: When different-sex couples are making personal decisions about their own relationships, they are not paying much attention to what same-sex couples are doing.

If any harm resulted from allowing same-sex couples to marry, it ought to be well documented by now. The fact that there has been no evidence of harms despite considerable effort to find some suggests that the predictions made by opponents of legalizing same-sex marriage were unwarranted at the time. Now that we have 20 years of research and experience, those predictions remain unwarranted now.

Benjamin Karney is a professor of social psychology at UCLA.

WASHINGTON — The Trump administration likes to promote its immigration enforcement agenda through numbers, with ambitious goals to deport 1 million people, report zero releases at the U.S.-Mexico border and arrest thousands of alleged gang members.

For all the boasting, the administration has been releasing less reliable, carefully vetted data than its predecessors on a signature policy that has become one of the most contentious of Trump’s second term.

The gap in information and a loss of figures from an office that has tracked immigration data back to the 1800s have left researchers, advocates, lawyers and journalists without important statistics to hold the Republican administration to account.

“They aren’t publishing the data,” said Mike Howell, who heads the conservative Oversight Project, an advocacy group pushing for more deportations. Instead, Howell said, the Department of Homeland Security has put out numbers in news releases “that purport to be statistics with no statistical backup and the numbers have jumped all over the place.”

With mass deportations a priority, new restrictions and increased enforcement have led to a surge in immigration arrests, detentions and deportations.

But finding the metrics that once measured those changes can be hard. It is an extension of earlier administration moves to limit the flow of government information by scrubbing or removing federal datasets or by the firing last year of the top official overseeing jobs data.

Important data is no longer publicly available

The Office of Homeland Security Statistics is responsible for publishing figures from Homeland Security agencies, including removals and the nationalities of those deported, to provide a comprehensive picture of immigration trends at the border and inside the United States.

Originally known as the Office of Immigration Statistics, it tracked such data since 1872. In its current form, created under the Biden administration, it also started publishing monthly reports that allowed researchers to track developments almost in real time.

But key enforcement metrics on its website have not been updated since early last year. A note on the page where the monthly reports were says it “is delayed while it is under review.”

“It’s the most timely data. It’s the most reliable data,” Austin Kocher, research professor at Syracuse University who closely follows immigration data trends, said about the monthly reports. “It has the most omniscient view of immigration enforcement across the entire agency.”

An interactive dashboard launched by U.S. Immigration and Customs Enforcement in December 2023 once let users examine whom the agency was arresting, their nationalities, criminal histories and removal numbers. ICE called it a “new era in transparency.”

Though intended for quarterly updates, the latest data is from January 2025. The agency’s annual report, typically released in December, had not been published as of mid-March.

Other agencies also publish data that touches on immigration, and parts of it do continue to roll out, such as U.S. Customs and Border Protection statistics detailing border encounters or data from the Department of Justice’s immigration courts.

But experts say other data has slowed.

The State Department’s most recent visa issuance data is from August. Key statistics from U.S. Citizenship and Immigration Services have not been updated since October.

The now-missing data had helped researchers study the effects of different policies. Lawyers could cite the figures to support their litigation. Journalists saw in them a powerful tool to hold the government to account on public claims or to report on important trends.

“We’re all a little bit in the dark about exactly how immigration enforcement is operating at a time when it’s taking new and unprecedented forms,” said Julia Gelatt, associate director of the U.S. Immigration Policy Program at the Migration Policy Institute.

DHS did not respond to detailed questions about why it was no longer releasing specific data.

“This is the most transparent Administration in history, we release new data multiple times a week and upon reporter request,” the department said in a statement.

Researchers contend with a patchwork of numbers

Figures the administration has released are inconsistent and unverifiable.

In a Jan. 20 news release, DHS said it had deported more than 675,000 people since Trump returned to the White House. A day later, in a second release, the department put the figure at 622,000. In congressional testimony March 4, Homeland Security Secretary Kristi Noem said the figure was 700,000.

But ICE, an agency within DHS, also releases figures on how many people it has removed from the country, part of a large data release mandated by Congress. An Associated Press analysis of the figures put that number at roughly 400,000 over Trump’s first year.

DHS has said 2.2 million people who were in the U.S. illegally have gone home on their own, but the department has given no explanation for the count. Experts have questioned the source of that figure, saying this was not something that DHS historically has tracked.

The department did not respond to questions about where that data came from.

With key sources of data halted, researchers, advocates and others have had to rely on information the administration is obliged to report or that has come to light through legal action.

The publication of ICE detention figures — how many people are detained, for how long and whether they have committed a crime — is required by Congress and is generally released every two weeks. But the figures’ release has faced some delays and its data gets overwritten with every new publication, complicating the work of people who need access to it.

The University of California, Berkeley’s Deportation Data Project, a research initiative, successfully sued through the Freedom of Information Act to access data about ICE arrests including nationalities, conviction status and whether arrests occurred at jails or in the community.

Graeme Blair, co-director of the project, said every administration has struggled with transparency in immigration enforcement, and given the Trump administration’s ambitious enforcement goals, the team wanted to secure and verify information that the government might not publicly release.

“Given the scale of what they were talking about doing, it seemed really important to be able to understand, to be able to double check those numbers,” he said.

But there are limitations, he said. The data obtained through the lawsuit only runs through Oct. 15. It does not cover recent operations such as the Minneapolis enforcement surge, when federal immigration officers fatally shot two protesters, leading to widespread demonstrations and scrutiny of enforcement tactics.

The absence of data is one of the few issues that has drawn bipartisan criticism.

“We deserve to know the numbers, just like we deserve to know who’s in our country and who needs to leave,” Howell said.

Santana writes for the Associated Press.