Glonce

WIRED's Uncanny Valley covers ICE's DNA dragnet and the growing backlash to AI slop

policyThe Glonce editorsWired🔥55Glonce rating55Impact50Novelty50Relevance60Surprise55Players60Credibility75
WIRED's Uncanny Valley covers ICE's DNA dragnet and the growing backlash to AI slop

This week's episode of WIRED's Uncanny Valley podcast, hosted by Brian Barrett, Zoe Schiffer and Leah Feiger, runs through several stories WIRED has been reporting on. The lead item is Immigration and Customs Enforcement's DNA collection: citing reporting by WIRED's Dhruv Mehrotra done in partnership with researchers at Georgetown, the hosts say ICE has collected DNA from nearly a million people this year, including young children with no criminal convictions, and that the Department of Homeland Security has become the single largest source of new profiles in an FBI database that keeps the samples indefinitely. Internal training materials the reporters obtained state that asylum seekers and refugees who have not adjusted their immigration status must have their DNA sampled, and that it must be collected after any arrest. WIRED and Mehrotra also identified two prosecutions filed in 2025 against people in immigration custody who refused to give a DNA sample. On the youngest cases, the hosts say the official line puts the collection age at 14 and up, but that Georgetown and WIRED found 492 children under 14 whose DNA the agency sent to the FBI between January 2025 and January 2026, including 21 five-year-olds, 32 six-year-olds and 33 seven-year-olds. Children were also swabbed at a family detention center in Dilley, Texas, prompting a group of lawmakers to say in a joint statement to WIRED that none of the families there have been convicted of a crime and do not belong in a database meant for violent criminals.

The hosts then turn to what they call AI slop. Google Earth rolled out a feature letting users type a prompt to superimpose an AI-generated scene onto real satellite imagery; within 24 hours people had used it to fabricate images including a nuclear plant in Iran, a bombed hospital in Gaza and fires at an Iranian oil terminal, and Google disabled the feature. The hosts frame this as part of a wider pattern of companies rolling back AI features in response to user pushback: LinkedIn added a button letting users flag a post as looking like AI slop, and Substack partnered with the AI-detection tool Pangram to let readers check whether a piece was potentially AI-generated. They also note that a new EU regulation took effect on Sunday requiring any AI-generated content encountered in a feed to be labeled as such. Schiffer adds that researchers she has spoken with at OpenAI, Anthropic and Google say they find it distinctive and irritating when their own executives post AI-written messages rather than writing their own.

Elsewhere in the episode, the hosts say the White House has announced a plan to address AI-related cybersecurity risks but is keeping most of the plan's details secret, and they discuss what that secrecy might mean for future regulation, without naming the officials or agency involved or giving a budget. They also mention that part of a SpaceX Falcon 9 rocket crashed into the moon, leaving a new crater, largely unnoticed at the time. The episode closes with WIRED senior writer Molly Taft joining to discuss her reporting on an unusual left-right political coalition that has formed to fight data centers across the US, a topic the intro flags but the available transcript does not detail further.

Key facts

  • ICE has collected DNA from nearly a million people this year, including young children, feeding an FBI database indefinitely; per WIRED and Georgetown researchers, 492 children under 14 had their DNA sent to the FBI between January 2025 and January 2026, including 21 five-year-olds, 32 six-year-olds and 33 seven-year-olds.
  • The Department of Homeland Security has become the single largest source of new profiles in the FBI's DNA database, and WIRED identified two 2025 prosecutions of people in immigration custody who refused DNA collection.
  • Users generated fabricated images, including a nuclear plant in Iran and a bombed hospital in Gaza, within 24 hours of the new Google Earth AI feature's launch, and Google quickly disabled it.
  • Backlash against AI-generated content is spreading: LinkedIn added a button to flag posts as AI slop, Substack partnered with the detector Pangram, and an EU regulation that took effect on Sunday now requires AI content in feeds to be labeled.
  • The White House is keeping most details of its new AI cybersecurity plan secret, and the hosts separately note a SpaceX Falcon 9 rocket stage crashed into the moon, largely unremarked at the time.

Why it matters

The episode is a snapshot of two trends moving in opposite directions at once. On one side, companies and regulators are visibly pulling back from AI: Google yanked its Earth AI feature after users exploited it within a day of launch, LinkedIn and Substack added ways for users to flag or detect AI content, and the EU's new labeling rule turns that instinct into law. On the other, a government agency is scaling up an AI-adjacent surveillance capability, DNA collection feeding a permanent FBI database, while the White House keeps its own AI security plan under wraps. The hosts read both as evidence of a widening gap between how Silicon Valley treats AI, as a productivity tool with real upside, and how much of the public experiences it, as something being imposed on them with little say or transparency.

Who it affects

Immigrants and asylum seekers, including children as young as five, whose DNA now sits in a permanent FBI database regardless of any criminal record. Google Earth users and anyone whose likeness or location could be manipulated by the AI feature before it was pulled. Creators and readers on LinkedIn and Substack, now able to flag or detect AI-generated posts, and anyone posting in the EU, who must now label AI content under the new rule. More broadly, anyone trying to gauge how seriously to take a White House cybersecurity plan whose contents are not public.

How to use it

The episode is available through the audio player on WIRED's site. To subscribe, iPhone and iPad users can open the Podcasts app or use WIRED's direct link; the show is also available through apps like Overcast or Pocket Casts by searching "uncanny valley," and on Spotify. The hosts can be followed on Bluesky: Brian Barrett at @brbarrett, Zoe Schiffer at @Zoeschiffer, Leah Feiger at @leahfeiger, and guest Molly Taft at @mollytaft.com. WIRED asks listeners to write in with questions or feedback.

How solid is it

The ICE segment rests on WIRED's own reporting by Dhruv Mehrotra, done in partnership with Georgetown researchers who obtained internal training materials, plus a direct joint statement from lawmakers quoted verbatim. The child-DNA figures and the count of 2025 prosecutions are attributed to that same WIRED and Georgetown work. Some framing, such as calling 14 "the official line" for DNA collection, is the hosts' own characterization rather than a cited document. WIRED notes the published transcript is automated and may contain errors, and the version captured for this summary trails off mid-sentence, so the SpaceX and data-center segments are described here only as introduced, not as discussed in full.

Risks and caveats

No calendar date is given for the podcast's release beyond "this week," for the EU regulation beyond "on Sunday," or for the SpaceX Falcon 9 lunar impact. The White House AI cybersecurity plan is described only as being kept largely secret; no officials, agency details or budget figures are given. The data-center politics segment with guest Molly Taft is flagged in the episode's intro but not covered in the transcript excerpt available here, so it is not detailed further in this summary.

“None of the families at Dilley have been convicted of a crime. They do not belong in a database meant for violent criminals, especially children.”

— a number of representatives, in a joint statement to WIRED