The Office of Personnel Management’s launch of a new data website earlier this month offered plenty more specifics on how the Trump administration reshaped the federal workforce over the last year.

OPM’s website depicts information and graphics on various changes the federal government saw throughout 2025, such as trends in overall staffing size, telework hours and much more.

But beyond that, the information OPM published contains further trends related to the federal workforce. The data, available through November 2025, provides even more insights into what exactly changed for agencies since President Donald Trump took office.

By total numbers, the agencies that saw the most employees leave their workforce — either voluntarily or involuntarily — were large departments like Defense and Agriculture.

With further analysis of OPM’s data, though, the trends show just how deeply federal employee separations impacted agencies. When broken down by how many employees left as a percentage of the total agency workforce, other agencies’ names emerge at the top of the list.

Notably, agencies that the Trump administration targeted throughout 2025, like USAID and the Education Department, saw much steeper levels of employees leaving, compared with some of the larger government departments. Although the federal employee separations were lower in number for these agencies, they represent much larger portions of their total workforce.

OPM’s data on federal employee separations covers those who quit their jobs, who were subject to a reduction in force (RIF), or who retired, as well as employees who transferred out of an agency, or who had a temporary position that expired sometime last year.

Regardless of the separation method, virtually all agencies saw significant federal employee separations throughout 2025, including all 24 CFO Act agencies.

Federal employee separations over the course of the year also varied from month to month. Depending on the time of year that employees officially separated from their jobs, clearer trends begin to emerge.

September was by far the highest month for federal workforce separations. More than 125,000 federal employees left their jobs in that month alone, coinciding with most agencies’ deadline for the deferred resignation program (DRP). Some agencies delayed their official DRP deadline to the end of December, but OPM data for that month is not yet available.

The types of jobs that saw the biggest numbers of employees leaving throughout 2025 also reveal key trends in the federal workforce data. Notably, governmentwide job series such as IT managers and HR managers are within the top 10 positions that saw employees exiting their jobs, according to OPM’s data.

Other job series like contact representatives also faced significant employee separations throughout the year, most notably at the IRS.

The post The federal workforce changed drastically in 2025. Here’s how. first appeared on Federal News Network.

© Federal News Network

Suspense, horror, wild action and dark humor make this a must-see film.

Sexual deepfakes continue to get more sophisticated, capable, easy to access, and perilous for millions of women who are abused with the technology.

Claude users will now be able to call up interactive apps inside the chatbot interface, with Cowork integration coming soon.

The 12-year-old firm raised $360,360,360 for its fifth fund. And there's a reason for that number.

The timing of these outages lines up with the long-awaited creation of a separate U.S. TikTok entity last week.

Europe-based venture fund 2150 raised the money to invest in climate tech startups that help solve cities' carbon footprint.

The London High Court awarded the London-based satirist and human rights activist Ghanem Al-Masarir more than £3 million, after finding the Saudi government hacked his phone and was likely behind a physical attack targeting him in London.

U.S. service members and their Italian counterparts gathered for the 82nd anniversary commemoration of Operation Shingle at the Sicily-Rome American Cemetery in Nettuno, Italy.

As a winter storm tightened its grip on Washington, National Guardsmen transformed their D.C. Safe and Beautiful mission from routine presence patrols toward life-safety priorities — mobility, visibility and helping people navigate snow and ice.

Malicious cyber actors may take advantage of weather crises. 

Prosecutors have been told they cannot rely on intercept evidence from the Sky ECC encrypted phone network in the absence of other evidence, unless they disclose the raw intercept data to defendants in decision that could have a significant impact on future prosecutions

Microsoft confirmed it can hand over BitLocker recovery keys stored in the cloud under warrant, reviving debate over who controls encrypted data.

The post Microsoft Shared BitLocker Keys With FBI, Raising Privacy Fears appeared first on TechRepublic.

In perhaps a matter of weeks, Bill Gates-backed TerraPower expects to receive federal permits to begin building the nuclear components of its first-of-a-kind, next-generation power plant in Wyoming.

The permits will put the company “a year ahead of anyone else,” predicted TerraPower CEO Chris Levesque. “We just keep proving that we’re the leading nuclear project in the U.S.”

After being largely mothballed for decades, America’s nuclear sector has kicked into hyper speed as tech giants scramble to power data centers nationwide and energy demands are spiking for commercial, residential and industrial uses.

TerraPower, which launched 20 years ago, aims to be the first in the U.S. to deliver on a new model of smaller, less expensive nuclear reactors that can be produced in three years from fabricated components — instead of the past approach of constructing giant, one-off structures that take a decade to erect.

The company is waiting for a green light from the U.S. Nuclear Regulatory Commission to break ground on the crucial next phase of its demonstration plant, whose construction in Kemmerer, Wyo., began in 2024. The goal is to start splitting atoms there by the end of 2030.

TerraPower’s Natrium technology features 345 megawatt nuclear devices that include a molten-salt thermal battery that increases output to 500 megawatts of power for limited periods.

The Bellevue, Wash.-based company has announced a string of new partnerships supporting its operations:

• It recently signed a deal with Meta to build up to eight advanced reactors in the U.S. with the first two coming online as soon as 2032. If the full order is fulfilled, all of the reactors aim to be operational by 2035. The companies are exploring multiple sites for the facilities.
• TerraPower shared last week that the electric utility giant Korea Hydro & Nuclear Power (KHNP) is now an investor, joining via a previously announced $250 million investment by SK, a South Korean multinational conglomerate.
• Last year, the company announced memorandums of understanding with government departments in Utah and Kansas to explore the potential siting of Natrium reactors in those states.
• TerraPower and HD Hyundai last March announced a collaboration in which the South Korean company will help manufacture components for the Natrium reactor.

“We plan to build hundreds of Natrium reactors,” Levesque told GeekWire. “We’re very focused on delivering the first one on time,” he added, and then quickly begin scaling.

A slate of challenges

TerraPower is competing against a slate of next-gen nuclear companies. Oklo was part of the announcement from Meta to support nuclear deployments, and could have a reactor operating as early as 2030 under the deal. Kairos Power is building a demonstration reactor in Tennessee in collaboration with Google that aims to come online that same year.

Last month, the U.S. Department of Energy awarded two grants supporting nuclear projects, one to the Tennessee Valley Authority to build a plant in that state and another to Holtec Government Services, which has a project in Michigan. Earlier this summer, X-energy and Amazon shared updates on a planned facility in Eastern Washington.

While TerraPower’s leadership is bullish on its prospects, there are plenty of hurdles ahead. They include limited supplies of reactor fuel, ongoing concerns about nuclear reactor safety, and construction cost challenges as designs evolve and a supply chain takes shape.

Finding fuel sources: Russia was the world’s only commercial producer of HALEU (high-assay, low-enriched uranium fuel, pronounced hay-lou), but that source was cut off after its 2022 invasion of Ukraine. Since then, the U.S. has restarted and been ramping up its domestic fuel production.

Levesque said the company will have what it needs for the Wyoming reactor, and is working with partners in South Africa and the U.S. to manufacture additional supplies.

Safety concerns: There are ongoing fears about the potential for a nuclear reactor to meltdown and release radioactive material, and concerns about the disposal and storage of spent fuel that remains radioactive for thousands of years.

At a recent public hearing in Olympia, Wash., regarding the potential for new nuclear facilities in the state, multiple participants cited worries about nuclear safety and environmental contamination.

Cost concerns: Research from Boston University analyzed nuclear plant construction costs going back decades, finding they were on average double the expected price tag, running nearly $2 billion over.

Levesque acknowledged the poor record, but said that as TerraPower goes into production mode, lessons learned from the demonstration project and each additional reactor will be incorporated to cut costs. He pointed to the SpaceX aerospace company as a model for that iterative approach.

But even if costs are controlled, some utility customers have expressed worries about who will pay for the deployment of the facilities and whether the projects will drive up electric bills.

The Wyoming plant was estimated in 2021 to cost about $4 billion; no updated figures have been provided. Levesque said private investments and a $2 billion federal grant mean that “we’re building that project without burdening the ratepayers.”

Pining for nuclear

And demand for nuclear has never been greater. “Almost any governor in the country now wants a nuclear plant,” Levesque said.

While wind and solar installations have expanded across America, electricity demand is projected to soar over the next years and decades. And while these renewable power sources are increasingly being paired with energy storage solutions such as giant batteries, nuclear is attractive for its 24/7 production capabilities.

Levesque said the company is being cautious in signing future deals, recognizing that it will be mid-2030s until it can hit its initial target of producing six reactors a year.

“We’re very careful not to over promise or over commit,” he said.

Microsoft on Monday announced Maia 200, the second generation of its custom AI chip, claiming it’s the most powerful first-party silicon from any major cloud provider. 

The company says Maia 200 delivers three times the performance of Amazon’s latest Trainium chip on certain benchmarks, and exceeds Google’s most recent tensor processing unit (TPU) on others.

The chip is already running workloads at Microsoft’s data center near Des Moines, Iowa. Microsoft says Maia 200 is powering OpenAI’s GPT-5.2 models, Microsoft 365 Copilot, and internal projects from its Superintelligence team. A second deployment at a data center near Phoenix is planned next.

It’s part of the larger trend among cloud giants to build their own custom silicon for AI rather than rely solely on Nvidia. Google has been refining its TPUs for nearly a decade, and Amazon’s Trainium line is now in its third generation, with a fourth already announced. 

Microsoft first revealed its custom chip ambitions in late 2023, when it unveiled Maia 100 at its Ignite conference. Despite entering the race late, Microsoft makes the case that its tight integration between chips, AI models, and applications like Copilot gives it an edge. 

The company says Maia 200 offers 30% better performance-per-dollar than its current hardware. Maia 200 also builds on the first-generation chip with a more specific focus on inference, the process of running AI models after they’ve been trained.

The chip competition among the cloud giants has intensified as the cost of running AI models becomes a bigger concern. Training a model is a one-time expense, but serving it to millions of users is a big ongoing expense. All three companies are betting that custom chips tuned for their own workloads will be cheaper than buying solely from Nvidia.

Microsoft is also opening the door to outside developers. The company announced a software development kit that will let AI startups and researchers optimize their models for Maia 200. Developers and academics can sign up for an early preview starting today.

Qwen:
Qwen releases Qwen3-Max-Thinking, its flagship reasoning model that it says demonstrates performance comparable to models such as GPT-5.2 Thinking and Opus 4.5  —  · QwenTeam丨Translations:.体中文  —  Introduction#  —  We present Qwen3-Max-Thinking, our latest flagship reasoning model.

Axios:
In a 38-page essay, Dario Amodei warns of civilization-level damage from superintelligent AI, questioning whether humanity has the maturity to handle such power  —  - Massive job loss: “I ... simultaneously think that AI will disrupt 50% of entry-level white-collar jobs over 1-5 years …

Angus Loten / Wall Street Journal:
Upwind Security, which filters out false security alerts in cloud apps to let security teams address actual risks, raised a $250M Series B at a ~$1.5B valuation  —  The startup aims to filter out a torrent of false alarms in cloud apps and workloads, enabling security teams to address actual risks

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

How do tech companies check if their users are kids?

This question has taken on new urgency recently thanks to growing concern about the dangers that can arise when children talk to AI chatbots. For years Big Tech asked for birthdays (that one could make up) to avoid violating child privacy laws, but they weren’t required to moderate content accordingly. Two developments over the last week show how quickly things are changing in the US and how this issue is becoming a new battleground, even among parents and child-safety advocates.

In one corner is the Republican Party, which has supported laws passed in several states that require sites with adult content to verify users’ ages. Critics say this provides cover to block anything deemed “harmful to minors,” which could include sex education. Other states, like California, are coming after AI companies with laws to protect kids who talk to chatbots (by requiring them to verify who’s a kid). Meanwhile, President Trump is attempting to keep AI regulation a national issue rather than allowing states to make their own rules. Support for various bills in Congress is constantly in flux.

So what might happen? The debate is quickly moving away from whether age verification is necessary and toward who will be responsible for it. This responsibility is a hot potato that no company wants to hold.

In a blog post last Tuesday, OpenAI revealed that it plans to roll out automatic age prediction. In short, the company will apply a model that uses factors like the time of day, among others, to predict whether a person chatting is under 18. For those identified as teens or children, ChatGPT will apply filters to “reduce exposure” to content like graphic violence or sexual role-play. YouTube launched something similar last year. 

If you support age verification but are concerned about privacy, this might sound like a win. But there’s a catch. The system is not perfect, of course, so it could classify a child as an adult or vice versa. People who are wrongly labeled under 18 can verify their identity by submitting a selfie or government ID to a company called Persona. 

Selfie verifications have issues: They fail more often for people of color and those with certain disabilities. Sameer Hinduja, who co-directs the Cyberbullying Research Center, says the fact that Persona will need to hold millions of government IDs and masses of biometric data is another weak point. “When those get breached, we’ve exposed massive populations all at once,” he says. 

Hinduja instead advocates for device-level verification, where a parent specifies a child’s age when setting up the child’s phone for the first time. This information is then kept on the device and shared securely with apps and websites. 

That’s more or less what Tim Cook, the CEO of Apple, recently lobbied US lawmakers to call for. Cook was fighting lawmakers who wanted to require app stores to verify ages, which would saddle Apple with lots of liability. 

More signals of where this is all headed will come on Wednesday, when the Federal Trade Commission—the agency that would be responsible for enforcing these new laws—is holding an all-day workshop on age verification. Apple’s head of government affairs, Nick Rossi, will be there. He’ll be joined by higher-ups in child safety at Google and Meta, as well as a company that specializes in marketing to children.

The FTC has become increasingly politicized under President Trump (his firing of the sole Democratic commissioner was struck down by a federal court, a decision that is now pending review by the US Supreme Court). In July, I wrote about signals that the agency is softening its stance toward AI companies. Indeed, in December, the FTC overturned a Biden-era ruling against an AI company that allowed people to flood the internet with fake product reviews, writing that it clashed with President Trump’s AI Action Plan.

Wednesday’s workshop may shed light on how partisan the FTC’s approach to age verification will be. Red states favor laws that require porn websites to verify ages (but critics warn this could be used to block a much wider range of content). Bethany Soye, a Republican state representative who is leading an effort to pass such a bill in her state of South Dakota, is scheduled to speak at the FTC meeting. The ACLU generally opposes laws requiring IDs to visit websites and has instead advocated for an expansion of existing parental controls.

While all this gets debated, though, AI has set the world of child safety on fire. We’re dealing with increased generation of child sexual abuse material, concerns (and lawsuits) about suicides and self-harm following chatbot conversations, and troubling evidence of kids’ forming attachments to AI companions. Colliding stances on privacy, politics, free expression, and surveillance will complicate any effort to find a solution. Write to me with your thoughts. 

Although there are a few hobbies that have low-cost entry points, amateur astronomy is not generally among them. A tabletop Dobsonian might cost a few hundred dollars, and that is just the entry point for an ever-increasing set of telescopes, mounts, trackers, lasers, and other pieces of equipment that it’s possible to build or buy. [Thomas] is deep into astronomy now, has a high-quality, remotely controllable telescope, and wanted to make it more accessible to his friends and others, so he built a system that lets the telescope stream on Twitch and lets his Twitch viewers control what it’s looking at.

The project began with overcoming the $4000 telescope’s practical limitations, most notably an annoyingly short Wi-Fi range and closed software. [Thomas] built a wireless bridge with a Raspberry Pi to extend connectivity, and then built a headless streaming system using OBS Studio inside a Proxmox container. This was a major hurdle as OBS doesn’t have particularly good support for headless operation.

The next step was reverse engineering the proprietary software the telescope uses for control. [Thomas] was able to probe network traffic on the Android app and uncovered undocumented REST and WebSocket APIs. From there, he gained full control over targeting, parking, initialization, and image capture. This allowed him to automate telescope behavior through Python scripts rather than relying on the official Android app.

To make the telescope interactive, he built a Twitch-integrated control system that enables viewers to vote on celestial targets, issue commands, and view live telemetry, including stacking progress, exposure data, and target coordinates. A custom HTML/CSS/JavaScript overlay displays real-time status, and there’s a custom loading screen when the telescope is moving to a new target. He also added ambient music and atmospheric effects, so the stream isn’t silent.

If [Thomas]’s stream is your first entry point into astronomy and you find that you need to explore it more on your own, there are plenty of paths to build your way into the hobby, especially with Dobsonian telescopes, which can be built by hand, including the mirrors.