Greetings everyone. Hope you all are hanging in there as everything continues to catch fire/freeze over. There’s been little meaningful fallout since the federal agents’ killing of Alex Pretti, other than the apparent dismissal of ICE figurehead Greg Bovino; operations are still continuing apace, Minneapolis is still in the streets, and Democrats abandoned the idea of using a shutdown threat as leverage. Tech workers do continue to call on executive leadership to condemn the administration and cancel ICE contracts—Silicon Valley CEOs have stayed quiet—and another day of work stoppage and protest is planned tomorrow, Friday, January 30th. NationalShutdown.org has a searchable map of actions across the US. I’ve received word that protestors are planning an ICE Out demonstration at an Amazon tech hub in Santa Monica. I’m planning on reporting on the event (otherwise, I’m observing the shutout) and I’ll include more details at the end of the post.

Despite the authoritarian headwinds, there are of course those working to keep democracy functioning. To that end, I was invited to speak at a meeting of California’s Legislative Progressive Caucus—a group of dozens of elected state assembly members—in San Francisco last week. The caucus is crafting its approach to its next legislative session, and I was asked to discuss AGI, labor, and tech policy in the age of Trump, along with co-panelists Emily Bender, the computational linguist at the University of Washington, computer scientist Margaret Mitchell, and AI researcher Deborah Raji. I thought I’d share my remarks with readers here, in case they’re of interest. I also wanted to note that I’m only able to do things like this thanks to my paying subscribers. Taking trips and preparing talks like this are entirely unpaid labor, and it’s only because I think it’s important to advocate for the worker, the human, and the user in the age of AI and big tech, that I undertake them. And subscribers make that work possible. Thank you.

Subscribe now

I’m also pleased to report that I was impressed by the defiant attitudes and outlook of many members of the progressive caucus I spoke with.Tthey made calls for showing up in the streets for their communities and protesting ICE alongside them. They laughed off the Trump administration’s efforts to ‘ban’ AI lawmaking via executive order. Shortly after the meeting, assemblyman Alex Lee, the head of the progressive caucus, introduced legislation to eliminate tax breaks for companies that contract with the ICE or the Department of Homeland Security. Isaac Bryan, who I’ve actually interviewed in these pages before, about his No Robo Bosses bill, sponsored legislation to ban police officers from taking secondary employment with ICE. Unlike too many Democrats at the federal level, they appeared to have no intention of standing down.

Hello -

Thanks for having me, and for convening this listening session. I’ve been been reporting on the tech industry for over 15 years, and for the last five, I’ve been focusing on technology and labor. I wrote a book called Blood in the Machine about the history of worker resistance to automation, and the perennially misunderstood Luddite uprising.

With that in mind, there are a few points I want to hit on:

First, that the history of automation technologies can teach us a great deal about the present. From the power loom to the robotized assembly line to artificial intelligence, he ways that automation technologies impact working people tend to follow similar patterns.

One key lesson is that stories about the incredible power of a new technology have historically been used by entrepreneurs, bosses, and elites to override norms, standards, and laws, or bend them to their favor. Factory owners who adopted mechanized looms in the early 1800s argued that the rules on the books shouldn’t apply to them because they were using new and improved technology. On those grounds, they eventually convinced British Parliament to tear up most of the laws protecting cloth workers altogether. Decades of immense working class suffering ensued.

Today, a lot of tech CEOs right down the street from here are arguing much the same thing, calling for a moratorium on AI laws and deregulation—arguing that AI technology is so new and powerful it shouldn’t be beholden to any state laws at all.

A key to this agenda is the idea that artificial general intelligence, or AGI, is about to arise and unleash untold economic and social benefits. This, too, I might note, is not new. Pro-business interests have been claiming that full automation is right around the corner for no less than two centuries. The inventor of the early computer, Charles Babbage, and the writer Andrew Ure notably did so in the 1800s, when they claimed autonomous, human-free factories were incipient—in part to rebut growing concerns over the brutal working conditions in real-world factories.

Now, generative AI is undoubtedly on many counts, novel, and technologically impressive. But I bring up all this historical context to remind us that we’ve been here before, with powerful interests using the specter of world-changing technology to concentrate power, justify trampling workers in the short term, and distract from the more immediate harms and impacts.

So that’s the second thing I want to underline here:

AGI, or an AI super intelligence should, in my opinion as a tech reporter and something of a historian of automation, be understood primarily as Emily Bender have described it: A marketing device. There are plenty of AI researchers who truly believe we are on the cusp of realizing an AI super intelligence, and there are plenty who do not. But figures like Sam Altman have realized that AGI makes for the ultimate story to sell partners and investors on his company and products.

OpenAI defines AGI as, and I quote from its official charter, still online today, “highly autonomous systems that outperform humans at most economically valuable work.” This is what it’s selling to enterprise clients around the world, and why so many companies have FOMO; no company wants to risk missing out on the ‘eliminate all labor costs’ machine. That’s ultimately why we have an AI bubble right now.

In fact, in researching a report for the AI Now Institute, I analyzed AI CEOs’ public usage of the term ‘AGI’, and found that it tended to correlate almost exactly to when they needed to raise a new funding round or distract from a PR scandal.

The third and last point I want to make is that, just as the dreams of the fully automated factory helped obscure real-world problems with factory conditions and child labor abuses in the 1800s, the AGI story is obscuring the fact that AI is causing a litany of real-world problems right now.

For one, it’s a motor behind the often reckless data center buildout that’s raising electricity bills, threatening water supplies, and impacting communities across the nation. But that’s all worth it, the AI companies say to states and municipalities eager to stay ahead of the curve, because AGI is around the bend.

Furthermore, many companies that buy AI firms’ enterprise software are buying into the AGI dream, too, and have indeed begun replacing workers’ jobs and tasks with AI, regardless of whether the technology is truly up for the task. So far, it’s usually not; the AI does a worse job than the human in nearly every case. But the AI is cheaper. So what you see happening a lot is firms executing layoffs—Salesforce, here in SF, for instance, cited AI as a driving force behind its most reason round of firings, as did Amazon recently—and then the remaining workers have to work extra hours to pick up the slack. In fact, after Amazon’s corporate leadership attributed its layoffs to AI, Amazon tech workers took the extraordinary step of issuing a public statement saying internal AI tools were not capable of doing their work, and that AI was an excuse to cut headcount.

This is happening in a number of different fields, not just tech. I run a series on my website called ‘AI Killed My Job’ in which I ask workers to share stories of how AI impacted their workplaces. I’ve heard from hundreds of workers at this point. And it is clear that AI *is* doing some real damage, especially to our creative industries here in California. Artists who’ve lost half of their clients to Midjourney and can no longer afford to pay rent. Translators who’ve anxiously watched their work dry up and are scrambling to change careers late in life. Copywriters who are instructed by bosses to feed their writing into a program so an AI model can train to become their replacement.

One of the stories I thought I’d shared here was one I received was from a man named Jacques, who was the copy chief for a tech company’s website and support pages. He oversaw all the documents that helped users understand how to use its product, and how to troubleshoot when something went wrong. He sent me this story in early 2025. I’ll quote from him:

“AI didn’t quite kill my current job, but it does mean that most of my job is now training AI to do a job I would have previously trained humans to do. I have no idea how entry-level developers, support agents, or copywriters are supposed to become senior devs, support managers, or marketers when the experience required to ascend is no longer available.”

I followed up with Jacques six months later, and he told me his company had laid him off. “I was actually let go the week before Thanksgiving now that the AI was good enough,” he wrote.

None of the above cases are happening, it’s important to underline, because there’s an AGI that is suddenly better than humans at telling stories, writing copy, creating art. People overwhelmingly prefer the work of humans to AI, in fact. But as with every significant automation technology past, AI is alluring for the simple reason that it lets bosses cut labor costs and exert more control over their workplaces.

The concern is that this idea of AGI is helping to cover for the rote, mass automation of lots of jobs I think we really want people to be able to do; translator, artist, actor, writer, therapists, nurses, I could go on — and if we allow it to continue apace, society will be all the poorer, more polluted, and beholden to systems controlled by a relative handful billionaires in Silicon Valley.

Industry interests will always try to use new automation technologies to justify undermining labor. It’s a recurring theme through automation history. They will tell stories about its power, its exceptionalism, to try to escape scrutiny and accountability for what’s actually happening. But there is absolutely no reason that any technology should not be subject to democracy, should not be shaped by the very people who must live with it. Technology should be built to benefit everyone—not just the powerful.

A final note: People are not only anxious about AI but angry about how it’s being forced on them in their workplaces, and what management is using it to do to their livelihoods. You’ve probably heard form your constituents. I just want to relay that I hear it from my interview subjects, from workers, students, and members of civil society, nearly every day. There is a lot of political power to be found in building solidarity against extractive AI companies, big tech and in strong action to curb their worse impulses—even, sometimes, refusing it.

Subscribe now

Amazon is a major ICE contractor. Last year, it inked a $25 million deal to supply the agency with cloud compute services. As such, it makes for a compelling setting for a protest during tomorrow’s ICE out actions in Santa Monica. Details are below:

That’s it for this week. Take the day off tomorrow, and stay safe out there. Hammers up.

Both of these statements are true:

1. Across the whole US economy, there’s not really a visible effect of AI on hiring and job mix;
2. Some sectors are absolutely devastated directly by AI.

But also:

1. Nobody cares if it was technically AI or not that took their job;
2. The wider economy is visibly screwed already.

Even the most mainstream financial press is starting to admit that claiming your layoffs are AI at work is a fake excuse to sound good to investors.

Here’s a headline from Fortune: “AI layoffs are looking more and more like corporate fiction that’s masking a darker reality, Oxford Economics suggests”. That darker reality is that the economy is already screwed. But we’ll get to that: [Fortune, archive]

The primary motivation for this rebranding of job cuts appears to be investor relations. The report notes that attributing staff reductions to AI adoption “conveys a more positive message to investors” than admitting to traditional business failures, such as weak consumer demand or “excessive hiring in the past.” By framing layoffs as a technological pivot, companies can present themselves as forward-thinking innovators rather than businesses struggling with cyclical downturns.

… When asked about the supposed link between AI and layoffs, Cappelli urged people to look closely at announcements. “The headline is, ‘It’s because of AI,’ but if you read what they actually say, they say, ‘We expect that AI will cover this work.’ Hadn’t done it. They’re just hoping. And they’re saying it because that’s what they think investors want to hear.”

A report from Yale’s Budget Lab says there isn’t evidence of economy-wide effects from AI: [FT, archive]

The labour market doesn’t feel great, so it feels correct that AI is taking people’s jobs. But we’ve looked at this many, many different ways, and we really cannot find any sign that this is happening.

Broadly, across economic sectors, AI isn’t visibly affecting the job market. But it’s a good layoff excuse: [CNBC]

Stephany said there isn’t much evidence from his research that shows large levels of technological unemployment due to AI.

“Economists call this structural unemployment, so the pie of work is not big enough for everybody anymore and so people will lose jobs definitely because of AI, I don’t think that this is happening on a mass scale,” he said.

So if that’s true, why are there all these layoffs? It’s broader and long-running economic problems. You can start at the end of the zero interest rate policy.

From 2007 to 2009, we had the global financial crisis. The US economy was so damaged by the crash, the Fed lowered interest rates to near-zero for most of a decade just to keep the money moving. You could borrow money almost free! So companies went as big as they could on the free money. They over-hired just in case they needed the workers.

Then in 2022, inflation hit and the Fed put interest rates back up. Suddenly, things were not going so great. Come to 2024–25, and companies are throwing out employees they hired prospectively like they’re surplus unsold stock.

So people are in fact losing their jobs. But don’t say “no-one is losing their jobs to AI.” That’s not actually true. Some sectors really have been devastated by AI specifically.

Brian Merchant at Blood in the Machine – which you should all read – has been hammering on this theme. He’s got an ongoing project, “AI Took My Job,” talking to workers who were indeed fired directly for AI.

Translators in particular — businesses think bad machine translation is good enough. Duolingo’s AI-generated content quality is so bad that a lot of paying customers have left. Freelance translators can hardly find work these days, specifically because chatbots mash out translations. [Blood in the Machine]

Content moderators are an AI target too, because companies really do not care about the job at all. They’ll do it with any old rubbish, and now they are! [Blood in the Machine]

The vibe-coding push and computer software really just not working any more, I don’t have a smoking gun link, but it’s clear that quality is job number 55 or so.

MBAs loathe employees. Any employees. They despise you. AI promises the one thing MBAs want more than anything — firing people — so they’re all-in. That it doesn’t work does not matter.

MBAs also assume any job they don’t understand must be simple, so they put out sloppy trash that any consumer can see is obviously terrible. But the product won’t lose its customer base for the next year — probably — so that’s a problem for several quarters from now.

Sometimes managers forget AI’s just the excuse, and they fire entry level workers assuming AI can replace them — when it can’t. Sometimes they realise they shot themselves in the foot. [Register]

What will happen is that companies will realise the bots can’t do the jobs. But this will take a year or two. Then the companies that survive will rehire people. They’ll try to do it at lower pay, of course. [Register]

Liz Fong-Jones at Honeycomb, and formerly of Google, says: [Bluesky]

AI today is literally not capable of replacing the senior engineers they are laying off. corps are in fact getting less done, but they’re banking on making an example of enough people that survivors put their heads down and help them implement AI in exchange for keeping their jobs … for now.

When the AI bubble pops — which I’m still guessing at next year — that will mark the start of Great Depression 2. It’s going to be bad.

But it’s bad already — without a few huge tech companies swapping the same 10 billion dollars around, the economy numbers officially go into recession.

But the real economy where you and I live is already in trouble. The numbers in the real US economy are so bad, President Trump fired the US Commissioner of Labor Statistics because of a jobs report he didn’t like. That’s how you know it’s going great! [BBC]

For now, it’s mutual aid time. If you’re still in work, send some money to the people who aren/t. They need it.

And, of course, we must mention our good friends at Stop Gen AI, who help redistribute money to those affected by AI-related, or AI-excuse, job cuts. Go support Stop Gen AI. [Stop Gen AI]

• Video — Podcast

It’s pledge week at Pivot to AI! If you enjoyed this post, and our other posts, please do put $5 into the Patreon. It helps us keep Pivot coming out daily. Thank you all.

A hacker demonstrated that the viral new AI agent Moltbot (formally Clawdbot) is easy to hack via a backdoor in an attached support shop.Clawdbot has become a Silicon Valley sensation among a certain type of AI-booster techbro, and the backdoor highlights just one of the things that can go awry if you use AI to automate your life and work.

Software engineer Peter Steinberger first released Moltbot as Clawdbot last November. (He changed the name on January 27 at the request of Anthropic who runs a chatbot called Claude.) Moltbot runs on a local server and, to hear its boosters tell it, works the way AI agents do in fiction. Users talk to it through a communication platform like Discord, Telegram, or Signal and the AI does various tasks for them.

According to its ardent admirers, Moltbot will clean up your inbox, buy stuff, and manage your calendar. With some tinkering, it’ll run on a Mac Mini and it seems to have a better memory than other AI agents. Moltbot’s fans say that this, finally, is the AI future companies like OpenAI and Anthropic have been promising. 

The popularity of Moltbot is sort of hard to explain if you’re not already tapped into a specific sect of Silicon Valley AI boosters. One benefit is the interface. Instead of going to a discrete website like ChatGPT, Moltbot users can talk to the AI through Telegram, Signal, or Teams. It’s also active, rather than passive. It also takes initiative. Unlike Claude or Copilot, Moltbot takes initiative and performs tasks it thinks a user wants done. The project has more than 100,000 stars on GitHub and is so popular it spiked Cloudflare’s stock price by 14% earlier this week because Moltbot runs on the service’s infrastructure.

But inviting an AI agent into your life comes with massive security risks. Hacker Jamieson O'Reilly demonstrated those risks in three experiments he wrote up as long posts on X. In the first, he showed that it’s possible for bad actors to access someone’s Moltbot through any of its processes connected to the public facing internet. From there, the hacker could use Moltbot to access everything else, including Signal messages, a user had turned over to Moltbot.

In the second post, O'Reilly created a supply chain attack on Moltbot through ClawdHub. “Think of it like your mobile app store for AI agent capabilities,” O’Reilly told 404 Media. “ClawdHub is where people share ‘skills,’ which are basically instruction packages that teach the AI how to do specific things. So if you want Clawd/Moltbot to post tweets for you, or go shopping on Amazon, there's a skill for that. The idea is that instead of everyone writing the same instructions from scratch, you download pre-made skills from people who've already figured it out.”

The problem, as O’Reilly pointed out, is that it’s easy for a hacker to create a “skill” for ClawdHub that contains malicious code. That code could gain access to whatever Moltbot sees and get up to all kinds of trouble on behalf of whoever created it.

For his experiment, O’Reilly released a “skill” on ClawdHub called “What Would Elon Do” that promised to help people think and make decisions like Elon Musk. Once the skill was integrated into people’s Moltbot and actually used, it sent a command line pop-up to the user that said “YOU JUST GOT PWNED (harmlessly.)”

Another vulnerability on ClawdHub was the way it communicated to users what skills were safe: it showed them how many times other people had downloaded it. O’Reilly was able to write a script that pumped “What Would Elon Do” up by 4,000 downloads and thus make it look safe and attractive. 

“When you compromise a supply chain, you're not asking victims to trust you, you're hijacking trust they've already placed in someone else,” he said. “That is, a developer or developers who've been publishing useful tools for years has built up credibility, download counts, stars, and a reputation. If you compromise their account or their distribution channel, you inherit all of that.”

In his third, and final, attack on Moltbot, O’Reilly was able to upload an SVG (vector graphics) file to ClawdHub’s servers and inject some JavaScript that ran on ClawdHub’s servers. O’Reilly used the access to play a song from The Matrix while lobsters danced around a Photoshopped picture of himself as Neo. “An SVG file just hijacked your entire session,” reads scrolling text at the top of a skill hosted on ClawdHub.

O’Reilly attacks on Moltbot and ClawdHub highlight a systemic security problem in AI agents. If you want these free agents doing tasks for you, they require a certain amount of access to your data and that access will always come with risks. I asked O’Reilly if this was a solvable problem and he told me that “solvable” isn't the right word. He prefers the word “manegeable.”

“If we're serious about it we can mitigate a lot. The fundamental tension is that AI agents are useful precisely because they have access to things. They need to read your files to help you code. They need credentials to deploy on your behalf. They need to execute commands to automate your workflow,” he said. “Every useful capability is also an attack surface. What we can do is build better permission models, better sandboxing, better auditing. Make it so compromises are contained rather than catastrophic.”

We’ve been here before. “The browser security model took decades to mature, and it's still not perfect,” O’Reilly said. “AI agents are at the ‘early days of the web’ stage where we're still figuring out what the equivalent of same-origin policy should even look like. It's solvable in the sense that we can make it much better. It's not solvable in the sense that there will always be a tradeoff between capability and risk.”

As AI agents grow in popularity and more people learn to use them, it’s important to return to first principles, he said. “Don't give the agent access to everything just because it's convenient,” O’Reilley said. “If it only needs to read code, don't give it write access to your production servers. Beyond that, treat your agent infrastructure like you'd treat any internet-facing service. Put it behind proper authentication, don't expose control interfaces to the public internet, audit what it has access to, and be skeptical of the supply chain. Don't just install the most popular skill without reading what it does. Check when it was last updated, who maintains it, what files it includes. Compartmentalise where possible. Run agent stuff in isolated environments. If it gets compromised, limit the blast radius.”

None of this is new, it’s how security and software have worked for a long time. “Every single vulnerability I found in this research, the proxy trust issues, the supply chain poisoning, the stored XSS, these have been plaguing traditional software for decades,” he said. “We've known about XSS since the late 90s. Supply chain attacks have been a documented threat vector for over a decade. Misconfigured authentication and exposed admin interfaces are as old as the web itself. Even seasoned developers overlook this stuff. They always have. Security gets deprioritised because it's invisible when it's working and only becomes visible when it fails.”

What’s different now is that AI has created a world where new people are using a tool they think will make them software engineers. People with little to no experience working a command line or playing with JSON are vibe coding complex systems without understanding how they work or what they’re building. “And I want to be clear—I'm fully supportive of this. More people building is a good thing. The democratisation of software development is genuinely exciting,” O’Reilly said. “But these new builders are going to need to learn security just as fast as they're learning to vibe code. You can't speedrun development and ignore the lessons we've spent twenty years learning the hard way.”

Moltbot’s Steinberger did not respond to 404 Media’s request for comment but O’Reilly said the developer’s been responsive and supportive as he’s red-teamed Moltbot. “He takes it seriously, no ego about it. Some maintainers get defensive when you report vulnerabilities, but Peter

immediately engaged, started pushing fixes, and has been collaborative throughout,” O’Reilly said. “I've submitted [pull requests] with fixes myself because I actually want this project to succeed. That's why I'm doing this publicly rather than just pointing my finger and laughing Ralph Wiggum style…the open source model works when people act in good faith, and Peter's doing exactly that.”

Acoustic solutions are essential in today’s workspaces, from pods to ceiling baffles, in all shapes and sizes. Most of these products, while effective, don’t always blend with interiors, and they often look more corporate than contemporary.

So when BIT CREATIVE was hired to envision a branch in Poland for leading Danish audio technology company GN Group, the architects decided to look at items in the core line, from hearing aids to devices for gamers. “The products were really the inspiration for the entire office concept,” says Barnaba Grzelecki, founder of BIT CREATIVE.

Located in the Saski Crescent complex in Warsaw, the 21,527-square-foot headquarters supports global operations, particularly the organization’s IT department, the largest cohort in the space. Exposed ceilings and tree canopies visible through some of the windows create an expansive feel.

Diverse zones allow employees to shift from heads-down work to group sessions with ease. An area for guests, which contains a plant-filled bookshelf, is reminiscent of a living room. The social sector at the heart of the workplace includes a casual dining section and bar. Glass blocks let sunlight filter in and complement the tile backsplash. There’s even a room dedicated to deep relaxation, complete with cosmic motifs and a recliner.

In a nod to Scandinavian style, natural oak wood is the material of choice, paired with red brick, concrete, and greenery. Found in the GN Group’s former office within a co-working center, these favored details now enhance the current setting. An LED ribbon that floats on the ceiling references sound waves, while graphics depict their intensity.

The color palette has a series of gray tones balanced by sage, navy, and hints of purple. The brand’s signature orange brings vibrancy to a music-themed room that features percussion instruments, including a display of cymbals.

Soundproofing is, of course, at the forefront, with multipurpose pieces that are ideal in the open office. Circular fixtures that resemble origami flowers seem to float overhead. PET panels integrated into partitions reduce noise and offer added privacy. Bitumen-backed carpet tiles bring durability and texture underfoot.

More than an office, the space doubles as a showroom-type environment that highlights sonic elements. “The design is built around sound, both functionally and visually,” notes Grzelecki.

To see this and other works by the studio, visit bitcreative.info.

Photography by Fotomohito.

Earlier this month we revealed Immigration and Customs Enforcement (ICE) is using a Palantir tool called ELITE to decide which neighborhoods to raid.

The tool lets ICE populate a map with potential deportation targets, bring up dossiers on each person, and view an address “confidence score” based on data sourced from the Department of Health and Human Services (HHS) and other government agencies. This is according to a user guide for ELITE 404 Media obtained.

404 Media is now publishing a version of that user guide so people can read it for themselves. 

404 Media is not publishing the original document for source protection reasons, but has retyped a version below. 404 Media has made some formatting tweaks for clarity and reconstructed some of the images in the guide (some of the images in the original are blurred). We previously did similar with an internal Palantir wiki in which the company explained its work with ICE.

Senator Ron Wyden previously told 404 Media in a statement, “The fact ICE is using this app proves the completely indiscriminate nature of the agency's aggressive and violent incursions into our communities. This app allows ICE to find the closest person to arrest and disappear, using government and commercial data, with the help of Palantir and Trump's Big Brother databases. It makes a mockery of the idea that ICE is trying to make our country safer. Rather, agents are reportedly picking people to deport from our country the same way you'd choose a nearby coffee shop.”

Palantir did not respond to a request for comment. But after publication of our original article, the company wrote a blog post, which said in part, “The ELITE tool is used for prioritized enforcement to surface the likely addresses of specific individuals, such as those with final orders of removal or with high severity criminal charges. The purpose of this tool is identifying the location of known foreign nationals who meet criteria for removal, not for mass prioritization of ‘locations where lots of people it might detain could be based.’”

One section of the guide indicates that during “special operations,” which target “groups of pre-defined aliens” leadership wants action against, normal safeguards in ELITE may need to be turned off. By default, the tool shows people with a final order of deportation, and with no reason preventing their removal (these can be turned off by a user of the tool). For special operations, the guide says those filters may need to be removed “to display all targets within a Special Operation dataset.” Palantir did not respond to a specific question asking if this undercuts the idea ELITE is a targeted, rather than broad, tool.

In testimony 404 Media obtained from a case in Oregon, an ICE official said ELITE is what ICE sometimes uses to track the apparent density of people at a particular location to target. “You’re going to go to a more dense population rather than [...] like, if there’s one pin at a house and the likelihood of them actually living there is like 10 percent [...] you’re not going to go there,” the official said. “It’s basically a map of the United States. It’s kind of like Google Maps.” That case related to a woman, who has the initials MJMA in the court records, who was detained with more than 30 other people in what her lawyers described as a dragnet. 

The Department of Homeland Security (DHS) did not respond to a request for comment.

The user guide follows below.

Overview

Enhanced Leads Identification & Targeting for Enforcement (ELITE) is a targeting tool designed to improve capabilities for identifying and prioritizing high-value targets through advanced analytics. Key functionality includes:

• ELITE enhances target viability and reduces officer workload by integrating new data sources

• ELITE employs an address confidence score that evaluates both the reliability of the source and the recency of the data to enhance prioritization

• ELITE provides the capability to export target lists

• ELITE auto-populates initial key FOW information

• ELITE interfaces with EID for encounters, EARM for cases, and EADM for detention data

• ELITE includes a geospatial heat map to better inform the field in developing articulable facts for consensual encounters

ELITE Workflow

The ELITE workflow includes all phases of targeting from identifying subjects to disposition.

1. Individuals are identified and sent to AOR-specific review queue

2. Officer conducts review of leads

3. Officer refines leads into list of actionable targets

4. Supervisors review the actionable targets and moves selections into the planning stage

5. After case is worked, officers disposition target

Home Screen

The ELITE home screen includes three main tabs: Enforcement Lead Tracker and Geospatial

Lead Sourcing, and UID Search.

The Enforcement Lead Tracker tab displays leads assigned to your queue, the Geospatial Lead Sourcing tab displays potential leads based on a map-view, and the UID Search tab allows you to look up individuals based on a Unique Identifier.

On the bottom left of the Home Screen, details about your profile are visible.

1. Creating a lead from UID Search

a. On the UlD tab you can search by many attributes for an individual anywhere within the immigration lifecycle.

NOTE: Ensure you verify entries and press Enter to run the search.

a. A list of results will appear in the Results pane.

b. Select a row to view individual details. Key Sections include:

i. Unique Identifiers: A-Number, FBI Number, etc. 

ii. Tags/Flags: Detention or criminal indicators. 

iii. Biographical Information. 

iv. Addresses.

c. On the top right corner, you may create a lead on this individual.

d. If the system detects an existing lead, it will be visible at the top of the pop-up.

e. Select a queue option (e.g., Enforcement Review Queue) to continue.

f. Click Create once the queue is selected

2. Creating bulk leads from UID Search

a. In the results pane, you can select up to 50 individuals with the checkboxes.

NOTE: Leads created in UID will populate into YOUR Enforcement Review Queue regardless of where the individual's highest confidence address is located.

Geospatial Lead Sourcing Tab

The Geospatial Lead Sourcing tab allows you to populate the Enforcement Review queue with leads by applying filters and selecting individuals from a map.

1. Apply Filters

a. ELITE default filters include Final Order= Yes, Reasons preventing removal= No, New Address Since Case Closed= No, and Active Case= Yes.

b. ELITE allows you to filter results by several other attributes organized by the following categories (Bio & IDs; Location; EID, EARM, IJ, CIS; Criminality (ACRIMe, NLETS, NCIC); Operations; Leads; Enrichments)

c. Confirm filters are correct and select View Results to see the cases displayed on the map

2. Select Individuals on the Map

a. You may select individuals one at a time or use the radius or polygon to select a focal area. You may need to zoom into your selection before drawing your focal area, then select individuals for review and possible nomination to the Enforcement Review queue. You are restricted to only viewing possible targets in your IMM current duty location.

3. Choose from Selected Individuals

a. Selected individuals will be listed on the bottom half of the screen.

b. Selecting an individual provides a preview of their record. Some of the data shown here includes name, A-Number, major case info, tags/flags for pertinent details, Unique IDs, and biographical data. The boxes highlighted across from the subject's Photo can be selected to show additional pertinent information. The individual timeline displays relevant encounters, detention, detainers and EOIR decisions on the right side of the screen.

NOTE: Address confidence score is a new feature of ELITE that factors in address recency and source (CLEAR, ACRIME Criminal Record, ACRIME RAP sheet Query, USCIS from UIP, HHS, NCATC, etc.) to generate a numerical value. Addresses are color coded (red, yellow, green). Red represents the address has a low confidence. Yellow represents a moderate confidence. Green represents a high confidence.

4. Create Leads

a. Selecting Create Leads will generate a pop-up with the headings New and Already Exists. Selecting the Create button continues the workflow by generating a new lead, but if a lead already exists it will be identified here for further review. Select Create to send the desired selected targets to the Enforcement Review queue/tab list. You can create up to 50 leads for review.

NOTE: A tagging feature enables potential leads to be labelled for easier organization and assigning for officer review.

Special Operations Filter

Special Operations are groups of pre-defined aliens specifically targeted by Leadership for action. Utilization of filters (including default filters) could change the intended target list for your area. As always, make sure you do your due diligence on each target to confirm removability prior to acting. Officers should consult ICE broadcasts or leadership for guidance on when to use these filters.

To filter by Special Operation, select the Click to Modify Filters button on the Geospatial Lead Sourcing tab.

The drop box will include a list of all available operations to select from

NOTE: Default filters such as Case Final Order Indicator = Yes and Are there reasons

preventing removal? = No may need to be removed to display all targets within a Special Operation dataset

Enforcement Review Queue

The Enforcement Review queue displays leads generated from the Geospatial Lead Sourcing tab. Here, officers are expected to validate leads in ELITE and determine which are best suitable to move to the Actionable Target queue.

1. In the Enforcement Review queue/tab the Apply Filters section can be used to isolate leads by office, keywords, or lead creator.

2. Selecting each alien individually will allow you to: Request Supervisor Action (reassignment or transfer) Add Enrichment or Note, Generate FOW, or Share.

a. Selecting Add Enrichment or Note will allow you to enter an Investigative Note, Attachment, Additional Address, Vehicle Information, Additional ID, Employer, Known Associate, Social Media, or Phone Number.

3. Select the check boxes on the desired aliens this will populate the Move Leads to Actionable Targets Queue section on the right. Select Submit to send the leads to the Actionable Targets queue.

NOTE: You can select each desired subject to review whether they are a good candidate for an actionable target list, and/or enrich the lead prior to sending it to the next queue. The Move to Actionable Targets Queue button can be used to send a single target to the next queue upon completion of a review/enrichment. The Archive Lead button moves the lead to a queue where it can still be enriched and is discoverable in the map, but it cannot be dispositioned.

NOTE: Archiving a lead will require the officer to enter a justification.

Tags

Tags serve to collate data for further operations based off user specified input, such as gang activity, local operation names or specific locations where operations will occur, as they relate to individuals. The Edit Tags button exists in the top selection bar within the enforcement lead tracker.

After selecting Edit Tags you will have the ability to create new tags, view details of existing tags and search for tags within your AOR. Below is an example of Create New Tag dialog box.

Users will be asked to create a Tag Name and to grant other users the ability to view or edit tags.

They also can also add leads to the tag upon creation.

NOTE: By default, all those in your AOR can view tags. Sharing further limits view to only those listed and tagging yourself limits the view to only yourself.

By clicking on View Details, users can further investigate information related to tags. Those with edit permission can also utilize the Edit Lead, Edit Viewers or Edit Editors button in the upper right corner. This will allow them to add or remove leads themselves, viewers or editors.

Users can also add Tag leads in bulk from within the Enforcement Review, Actionable Targets or Planning and Dispositioning queues. To do this, select the check box next to the leads you wish to tag, then select the Bulk Actions button in the upper right-hand corner. You will be given a preview pane and the ability to select from a drop down of tags you have access to within your AOR.

Actionable Targets Queue

The Actionable Targets queue displays a list of vetted targets deemed eligible for planning and execution. A supervisor is required to move a target from the Actionable Targets queue to the Planning queue.

(Supervisors only) to send leads to the Planning queue:

1. From the Actionable Targets queue, select the check boxes of the desired aliens on the list. You can send up to 15 leads for action.
2. Select Submit to send the leads to the Planning queue.

Planning Queue

Once in the Planning Queue, both officer and supervisors can continue to work the lead.

To create an operation packet:

1. Select the desired subjects and right-click anywhere to export the list to Excel

2. If desired, select each target individually that will be part of the operation and print the FOW.

Dispositioning

In the Dispositioning queue, officers can review leads post enforcement operations and provide an outcome/disposition status relative to each. The disposition status of the lead post operation is crucial for reporting purposes.

1. From the Planning queue, select the check boxes of the desired aliens on the list. You can send up to 15 leads for dispositioning.
2. Select Submit to send the leads to the Disposition queue.
3. From the Dispositioning queue, select each alien and select the appropriate disposition status from the dropdown. Disposition each alien individually.

Chat & Ask AI, one of the most popular AI apps on the Google Play and Apple App stores that claims more than 50 million users, left hundreds of millions of those users’ private messages with the app’s chatbot exposed, according to an independent security researcher and emails viewed by 404 Media. The exposed chats showed users asked the app “How do I painlessly kill myself,” to write suicide notes, “how to make meth,” and how to hack various apps. 

The exposed data was discovered by an independent security researcher who goes by Harry. The issue is a misconfiguration in the app’s usage of the mobile app development platform Google Firebase, which by default makes it easy for anyone to make themselves an “authenticated” user who can access the app’s backend storage where in many instances user data is stored. Harry said that he had access to 300 million messages from more than 25 million users in the exposed database, and that he extracted and analyzed a sample of 60,000 users and a million messages. The database contained user files with a complete history of their chats with the AI, timestamps of those chats, the name they gave the app’s chatbot, how they configured the model, and which specific model they used. Chat & Ask AI is a “wrapper” that plugs into various large language models from bigger companies users can choose from, Including OpenAI’s ChatGPT, Anthropic's Claude, and Google’s Gemini.