ICE Is Paying Palantir Million to Build ‘ImmigrationOS’ Surveillance Platform

ICE Is Paying Palantir $30 Million to Build ‘ImmigrationOS’ Surveillance Platform

AI & ML


“No other vendor could meet these timeframes of having the infrastructure in place to meet this urgent requirement and deliver a prototype in less than six months,” ICE says in the document.

ICE’s document does not specify the data sources Palantir would pull from to power ImmigrationOS. However, it says that Palantir could “configure” the case management system that it has provided to ICE since 2014.

Palantir has done work at various other government agencies as early as 2007. Aside from ICE, it’s worked with the US Army, Air Force, Navy, Internal Revenue Service, and Federal Bureau of Investigation. As reported by WIRED, Palantir is currently helping Elon Musk’s so-called Department of Government Efficiency (DOGE) build a brand new “mega API” at the IRS that could search for records across all the different databases that the agency maintains.

Last week, 404 Media reported that a recent version of Palantir’s case management system for ICE allows agents to search for people based on “hundreds of different, highly specific categories,” including how a person entered the country, their current legal status, and their country of origin. It also includes a person’s hair and eye color, whether they have scars or tattoos, and their license plate reader data, which would provide detailed location data about where that person travels by car.

These functionalities have been mentioned in a government privacy assessment published in 2016, and it’s not clear what new information may have been integrated into the case management system over the past four years.

This week’s $30 million award is an addition to an existing Palantir contract penned in 2022, originally worth about $17 million, for work on ICE’s case management system. The agency has increased the value of the contract five times prior to this month; the largest was a $19 million increase in September 2023.

The contract’s ImmigrationOS update was first documented on April 11 in a government-run database tracking federal spending. The entry had a 248-character description of the change. The five-page document ICE published Thursday, meanwhile, has a more detailed description of Palantir’s expected services for the agency.

The contract update comes as the Trump administration deputizes ICE and other government agencies to drastically escalate the tactics and scale of deportations from the US. In recent weeks, immigration authorities have arrested and detained people with student visas and green cards, and deported at least 238 people to a brutal megaprison in El Salvador, some of whom have not been able to speak with a lawyer or have due process.

As part of its efforts to push people to self-deport, DHS in late March revoked the temporary parole of more than half a million people and demanded that they self-deport in about a month, despite having been granted authorization to live in the US after fleeing dangerous or unstable situations in Cuba, Haiti, Nicaragua, and Venezuela under the so-called “CHNV parole programs.”

Last week, the Social Security Administration listed more than 6,000 of these people as dead, a tactic meant to end their financial lives. DHS, meanwhile, sent emails to an unknown number of people declaring that their parole had been revoked, and demanding them to self-deport. Several US citizens, including immigration attorneys, received the email.

On Monday, a federal judge temporarily blocked the Trump administration’s move to revoke the people’s authorization to live in the US under the CHNV programs. White House spokesperson Karoline Leavitt called the judge’s ruling “rogue.”



Source link

This ‘College Protester’ Isn’t Real. It’s an AI-Powered Undercover Bot for Cops

This ‘College Protester’ Isn’t Real. It’s an AI-Powered Undercover Bot for Cops

AI & ML


On June 5, a Pinal County Board of Supervisors meeting was asked to approve a $500,000 contract between the county and Massive Blue in order to license Overwatch.

“I was looking at the website for Massive Blue, and it’s a one-pager with no additional information and no links,” Kevin Cavanaugh, the then-supervisor for District 1, said to Pinal County’s Chief Deputy at the Sheriff’s Office, Matthew Thomas. “They produce software that we buy, and it does what? Can you explain that to us?”

“I can’t get into great detail because it’s essentially trade secrets, and I don’t want to tip our hand to the bad guys,” Thomas said. “But what I can tell you is that the software is designed to help our investigators look for and find and build a case on human trafficking, drug trafficking, and gun trafficking.”

Cavanaugh said at the board meeting that the basic information he got is that Massive Blue uses “50 AI bots.” He then asked whether the software has been successful and if it helped law enforcement make any arrests. Thomas explained they have not made any arrests yet because they’ve only seen the proof of concept, but that the proof of concept was “good enough for us and our investigators to move forward with this. Once this gets approved and we get them [Massive Blue] under contract, then we are going to move forward with prosecution of cases.”

Cavanaugh asked if Overwatch is used in other counties, which prompted Thomas to invite Clem to the podium to speak. Clem introduced himself as a recently retired border agent and said that Massive Blue is currently in negotiations with three counties in Arizona, including Pinal County.

“As a resident of 14 years of Pinal County I know what’s happening here,” Clem said to the Board of Supervisors. “To be able [to] use this program […] to provide all the necessary information to go after the online exploitation of children, trafficking victims, and all the other verticals that the sheriff may want to go after.”

Cavanaugh again asked if Massive Blue gathered any data that led to arrests.

“We have not made arrests yet, but there is a current investigation right now regarding arson, and we got the leads to the investigators,” Clem said, explaining that the program has been active for only about six months. “Investigations take time, but we’ve been able to generate the necessary leads for the particular counties that we’re involved with and also in the private sector.”

The Pinal County Board of Supervisors concluded the exchange by approving payment for a handful of other, unrelated projects, but with board members asking to delay the vote on payment for Massive Blue “for further study.”

The decision not to fund Massive Blue that day was covered in a local newspaper. Cavanaugh told the paper that he asked the company to meet with supervisors to explain the merits of the software.

“The State of Arizona has provided a grant, but grant money is taxpayer money. No matter the source of the funding, fighting human and sex trafficking is too important to risk half a million dollars on unproven technology,” he said. “If the company demonstrates that it can deliver evidence to arrest human traffickers, it may be worthwhile. However, it has yet to achieve this goal.”



Source link

If I Don’t Use AI, Will My Grandkids Still Think I’m Cool?

If I Don’t Use AI, Will My Grandkids Still Think I’m Cool?

AI & ML


As a retiree, I want to stay close to my grandkids. I worry that not learning how to use AI will leave me behind. What’s the easiest tool for me to learn, and should I be worried? —Lifelong Learner

Be not afraid! I promise that you do not need to learn how to use a generative AI tool like ChatGPT or Claude to ensure your grandkids see you as a relevant, informed person. If anything, I would say that our culture has tipped over the past year to generally oppose the use of generative AI tools due to their outsize environmental impact, ethical concerns over their data scraping, and general sludginess of the outputs. So, depending on your relatives’ worldview, confessing that you don’t use any chatbot tools might even boost your cool factor with them.

Also, AI chatbots are possibly eroding our social skills. OpenAI recently conducted a research study to determine whether repetitive ChatGPT usage made users feel lonelier and less social overall. It did; the most frequent ChatGPT users in the study became emotionally dependent on the chatbot’s company. So, if you want to spend time fostering better connections with your family, generative AI should be near the bottom of your list of essential tools to learn.

Instead, stay up to speed on your basic software-enabled communication skills like texting, emailing, video calling, and the proper uses of social media. Keeping these skills sharp may help you feel more connected online, both through direct communication with your family members and your more complete awareness of what the grandkids are doing day to day.

If you want to experiment with AI tools, I’d recommend using them with your family members for education and entertainment. When ChatGPT first came out in 2022 and I was visiting my parents for the holidays, we had a wonderful time trying it out together, making funny poems, and refining prompts just to see what the AI could do. If your grandkids are really into using this type of software, then asking if they could spend 30 minutes the next time you’re together in person playing around with the tool could be an enjoyable bonding experience.

If you’re not the fastest typer, you may find the voice options for inputting prompts into the chatbot more comfortable to use. ChatGPT’s voice mode is entertaining to chat with, though it will likely feel quite odd at first. I mean, how do you strike up a conversation with a piece of software? But once you get used to voice mode, it can be fun to ask it to teach you phrases in different languages. Before traveling to Japan last year, I spent hours and hours practicing speaking Japanese with ChatGPT. It’s not a perfect tutor, but I found the process to be quite enjoyable.

One thing to keep in mind is that chatbots often fabricate information or spit out search results that contain inaccuracies. This can happen when you’re doing web searches to gather specific information, or when you ask the chatbot a general question. The wrong answers given by these generative AI tools can sound quite confident, so you’re better off doing some additional digging to verify what it told you. When you’re doing this double-checking, you may realize there’s not much you can get from ChatGPT today that you couldn’t find through a simple Google search.

Another detail worth remembering: It’s not alive. Even if you feel impressed with the quality of the answers or get an uncanny feeling that it’s more than just a software program talking to you, remind yourself that chatbots are massive tools that memorize the patterns of human writing and interactions with the intention of mimicking them. The mimicry may be stunning, but the expressions are ultimately hollow.

Let’s get back to your worries about staying connected to family. Take a moment and genuinely think about your relationship with your grandkids. They hopefully love you for who you are, and what kind of software you use or avoid shouldn’t have any impact on that. I feel like life is far too short to be caught up in our anxieties about how others view us. As soon as you finish this article, give them a call (or send them an Instagram DM) and tell them how you feel as well as how important family is to you. Do it now. I promise it’s worth it. Besides, you already know what to say to them—no chatbot needed here.



Source link

Meet The AI Agent With Multiple Personalities

Meet The AI Agent With Multiple Personalities

AI & ML


In the coming years, agents are widely expected to take over more and more chores on behalf of humans, including using computers and smartphones. For now, though, they’re too error prone to be much use.

A new agent called S2, created by the startup Simular AI, combines frontier models with models specialized for using computers. The agent achieves state-of-the-art performance on tasks like using apps and manipulating files—and suggests that turning to different models in different situations may help agents advance.

“Computer-using agents are different from large language models and different from coding,” says Ang Li, cofounder and CEO of Simular. “It’s a different type of problem.”

In Simular’s approach, a powerful general-purpose AI model, like OpenAI’s GPT-4o or Anthropic’s Claude 3.7, is used to reason about how best to complete the task at hand—while smaller open source models step in for tasks like interpreting web pages.

Li, who was a researcher at Google DeepMind before founding Simular in 2023, explains that large language models excel at planning but aren’t as good at recognizing the elements of a graphical user interface.

S2 is designed to learn from experience with an external memory module that records actions and user feedback and uses those recordings to improve future actions.

On particularly complex tasks, S2 performs better than any other model on OSWorld, a benchmark that measures an agent’s ability to use a computer operating system.

For example, S2 can complete 34.5 percent of tasks that involve 50 steps, beating OpenAI’s Operator, which can complete 32 percent. Similarly, S2 scores 50 percent on AndroidWorld, a benchmark for smartphone-using agents, while the next best agent scores 46 percent.

Victor Zhong, a computer scientist at the University of Waterloo in Canada and one of the creators of OSWorld, believes that future big AI models may incorporate training data that helps them understand the visual world and make sense of graphical user interfaces.

“This will help agents navigate GUIs with much higher precision,” Zhong says. “I think in the meantime, before such fundamental breakthroughs, state-of-the-art systems will resemble Simular in that they combine multiple models to patch the limitations of single models.”

To prepare for this column, I used Simular to book flights and scour Amazon for deals, and it seemed better than some of the open source agents I tried last year, including AutoGen and vimGPT.

But even the smartest AI agents are, it seems, still troubled by edge cases and occasionally exhibit odd behavior. In one instance, when I asked S2 to help find contact information for the researchers behind OSWorld, the agent got stuck in a loop hopping between the project page and the login for OSWorld’s Discord.

OSWorld’s benchmarks show why agents remain more hype than reality for now. While humans can complete 72 percent of OSWorld tasks, agents are foiled 38 percent of the time on complex tasks. That said, when the benchmark was introduced in April 2024, the best agent could complete only 12 percent of the tasks.



Source link

There’s AI Inside Windows Paint and Notepad Now. Here’s How to Use It

There’s AI Inside Windows Paint and Notepad Now. Here’s How to Use It

AI & ML


Note that while Generative Erase and Remove Background can be used for free, Image Creator uses up AI credits associated with your Microsoft account. You can’t buy these separately, they come as part of a subscription to Microsoft 365 or Copilot Pro subscriptions, so use them wisely. You can read more about AI credits and how they work here.

AI in Windows Notepad

Get some Copilot help with your compositions in Notepad.

Courtesy of David Nield

Notepad is perhaps better known as a code editor than a word processor, but in recent years Microsoft has added more features in the way of formatting and auto-save. If you open it in Windows, you’ll see these features as well as a Copilot button in the top-right corner of the interface.

You can’t use Copilot inside Notepad to generate new text, as you can in Copilot on the web or in other tools like ChatGPT. Instead, the feature lets you rewrite and tweak what you’ve already written—so before you click on the Copilot button, you need to put some text into Notepad and then select it.

With the selection made, click the Copilot button, and you get a range of options: Make shorter and Make longer can obviously be used to change the length of the selected text, and you’ve also got a Change tone option if you want to make the text more inspirational, formal, casual, or humorous. There’s also Change format, which lets you put the selected text into a different structure: A list, marketing speak, or poetry, for example.

You can also choose Rewrite from this menu for a more comprehensive set of options—and to see previews of the rewritten text before it’s applied. A new pop-up window appears, giving you more options for changing the length, tone, and format. You also get different variations to choose between in each case. When you find something you like, click Replace to swap it out for the existing text.

At the time of writing, it seems Notepad is giving everyone a few AI-powered rewrites for free—but as with Image Creator in Paint, you’re going to need some AI credits with a Microsoft 365 or Copilot Pro subscription to use this extensively.

Of course, if you’d rather not use these AI tools and don’t want to see the Copilot button hanging around, you can turn it off altogether: Click the gear icon (top right), then turn off the Copilot toggle switch.



Source link