Since the middle of last year, there have been at least three major AI “acqui-hires” in Silicon Valley. Meta invested more than $14 billion in Scale AI and brought on its CEO, Alexandr Wang; Google spent a cool $2.4 billion to license Windsurf’s technology and fold its cofounders and research teams into DeepMind; and Nvidia wagered $20 billion on Groq’s inference technology and hired its CEO and other staffers.

The frontier AI labs, meanwhile, have been playing a high stakes and seemingly never-ending game of talent musical chairs. The latest reshuffle began three weeks ago, when OpenAI announced it was rehiring several researchers who had departed less than two years earlier to join Mira Murati’s startup, Thinking Machines. At the same time, Anthropic, which was itself founded by former OpenAI staffers, has been poaching talent from the ChatGPT maker. OpenAI, in turn, just hired a former Anthropic safety researcher to be its “head of preparedness.”

The hiring churn happening in Silicon Valley represents the “great unbundling” of the tech startup, as Dave Munichiello, an investor at GV, put it. In earlier eras, tech founders and their first employees often stayed onboard until either the lights went out or there was a major liquidity event. But in today’s market, where generative AI startups are growing rapidly, equipped with plenty of capital, and prized especially for the strength of their research talent, “you invest in a startup knowing it could be broken up,” Munichiello told me.

Early founders and researchers at the buzziest AI startups are bouncing around to different companies for a range of reasons. A big incentive for many, of course, is money. Last year Meta was reportedly offering top AI researchers compensation packages in the tens or hundreds of millions of dollars, offering them not just access to cutting-edge computing resources but also … generational wealth.

But it’s not all about getting rich. Broader cultural shifts that rocked the tech industry in recent years have made some workers worried about committing to one company or institution for too long, says Sayash Kapoor, a computer science researcher at Princeton University and a senior fellow at Mozilla. Employers used to safely assume that workers would stay at least until the four-year mark when their stock options were typically scheduled to vest. In the high-minded era of the 2000s and 2010s, plenty of early cofounders and employees also sincerely believed in the stated missions of their companies and wanted to be there to help achieve them.

Now, Kapoor says, “people understand the limitations of the institutions they’re working in, and founders are more pragmatic.” The founders of Windsurf, for example, may have calculated their impact could be larger at a place like Google that has lots of resources, Kapoor says. He adds that a similar shift is happening within academia. Over the past five years, Kapoor says, he’s seen more PhD researchers leave their computer-science doctoral programs to take jobs in industry. There are higher opportunity costs associated with staying in one place at a time when AI innovation is rapidly accelerating, he says.

Investors, wary of becoming collateral damage in the AI talent wars, are taking steps to protect themselves. Max Gazor, the founder of Striker Venture Partners, says his team is vetting founding teams “for chemistry and cohesion more than ever.” Gazor says it’s also increasingly common for deals to include “protective provisions that require board consent for material IP licensing or similar scenarios.”

Gazor notes that some of the biggest acqui-hire deals that have happened recently involved startups founded long before the current generative AI boom. Scale AI, for example, was founded in 2016, a time when the kind of deal Wang negotiated with Meta would have been unfathomable to many. Now, however, these potential outcomes might be considered in early term sheets and “constructively managed,” Gazor explains.

The face-recognition app Mobile Fortify, now used by United States immigration agents in towns and cities across the US, is not designed to reliably identify people in the streets and was deployed without the scrutiny that has historically governed the rollout of technologies that impact people’s privacy, according to records reviewed by WIRED.

The Department of Homeland Security launched Mobile Fortify in the spring of 2025 to “determine or verify” the identities of individuals stopped or detained by DHS officers during federal operations, records show. DHS explicitly linked the rollout to an executive order, signed by President Donald Trump on his first day in office, which called for a “total and efficient” crackdown on undocumented immigrants through the use of expedited removals, expanded detention, and funding pressure on states, among other tactics.

Despite DHS repeatedly framing Mobile Fortify as a tool for identifying people through facial recognition, however, the app does not actually “verify” the identities of people stopped by federal immigration agents—a well-known limitation of the technology and a function of how Mobile Fortify is designed and used.

“Every manufacturer of this technology, every police department with a policy makes very clear that face recognition technology is not capable of providing a positive identification, that it makes mistakes, and that it’s only for generating leads,” says Nathan Wessler, deputy director of the American Civil Liberties Union’s Speech, Privacy, and Technology Project.

Records reviewed by WIRED also show that DHS’s hasty approval of Fortify last May was enabled by dismantling centralized privacy reviews and quietly removing department-wide limits on facial recognition—changes overseen by a former Heritage Foundation lawyer and Project 2025 contributor, who now serves in a senior DHS privacy role.

DHS—which has declined to detail the methods and tools that agents are using, despite repeated calls from oversight officials and nonprofit privacy watchdogs—has used Mobile Fortify to scan the faces not only of “targeted individuals,” but also people later confirmed to be US citizens and others who were observing or protesting enforcement activity.

Reporting has documented federal agents telling citizens they were being recorded with facial recognition and that their faces would be added to a database without consent. Other accounts describe agents treating accent, perceived ethnicity, or skin color as a basis to escalate encounters—then using face scanning as the next step once a stop is underway. Together, the cases illustrate a broader shift in DHS enforcement toward low-level street encounters followed by biometric capture like face scans, with limited transparency around the tool’s operation and use.

Fortify’s technology mobilizes facial capture hundreds of miles from the US border, allowing DHS to generate nonconsensual face prints of people who, “it is conceivable,” DHS’s Privacy Office says, are “US citizens or lawful permanent residents.” As with the circumstances surrounding its deployment to agents with Customs and Border Protection and Immigration and Customs Enforcement, Fortify’s functionality is visible mainly today through court filings and sworn agent testimony.

In a federal lawsuit this month, attorneys for the State of Illinois and the City of Chicago said the app had been used “in the field over 100,000 times” since launch.

In Oregon testimony last year, an agent said two photos of a woman in custody taken with his face-recognition app produced different identities. The woman was handcuffed and looking downward, the agent said, prompting him to physically reposition her to obtain the first image. The movement, he testified, caused her to yelp in pain. The app returned a name and photo of a woman named Maria; a match that the agent rated “a maybe.”

Agents called out the name, “Maria, Maria,” to gauge her reaction. When she failed to respond, they took another photo. The agent testified the second result was “possible,” but added, “I don’t know.” Asked what supported probable cause, the agent cited the woman speaking Spanish, her presence with others who appeared to be noncitizens, and a “possible match” via facial recognition. The agent testified that the app did not indicate how confident the system was in a match. “It’s just an image, your honor. You have to look at the eyes and the nose and the mouth and the lips.”

However, rendered here in Motorola’s Watch app, everything looks fun and easy! Motorola (and Polar, I guess) uses Apple’s “close your rings” approach, with active minutes, steps, and calories. I particularly like that you can now use Polar’s sleep tracking with a cheaper Android watch. Polar takes into account sleep time, solidity (whether or not your sleep was interrupted), and regeneration to give you a Nightly Recharge Status.

You can still click through and see your ANS, but there’s a lot more context surrounding it. Also, the graphs are prettier. I compared the sleep, heart rate, and stress measurements to my Oura Ring 4, and I found no big discrepancies. The Moto Watch tended to be a little bit more generous in my sleep and activity measurements (7 hours and 21 minutes of sleep instead of 7 hours and 13 minutes, or 3,807 steps as compared to 3,209), but that’s usual for lower-end fitness trackers that have fewer and less-sensitive sensors.

On that note, I do have one major hardware gripe. Onboard GPS is meant to make it easier to just run out the door and start your watch. I didn’t find this to be the case. Whatever processor is in the watch (Motorola has conveniently chosen not to reveal this), it’s just really slow to connect to satellites and iffy whenever it does. This isn’t a huge deal when I’m just walking my dog or lifting weights in my living room, but it constantly cuts out when I’m outside and doesn’t have the ability to fill in the blanks, as another, more expensive fitness tracker would do.

It’s just really annoying to constantly get pinged about satellite loss and to have a quarter-mile or a half-mile cut out of your runs. That’s how I know the speaker works—it was constantly telling me it lost satellite connection during activities.

Finally, the screen and buttons are really sensitive. It does give you an option to lock the screen, but even then, I found myself accidentally unlocking it from time to time and turning the recording off when I didn’t mean to.

As I write this, I have seven different smartwatches from different brands sitting on my desk. If you’re looking for a cheap, attractive, and effective Android-compatible smartwatch, I would say that the CMF Watch 3 Pro is your best choice. However, I do think the integration with Polar was well done, and the price point is not that bad. I’m definitely keeping an eye out for what Motorola might have to offer in the future.