I gave CNET Editors’ Choice awards to Baseus’ Bose-infused Baseus Inspire XH1 headphones and Inspire XP1 earbuds because they’re well designed and sound decent consider their prices. I also liked Baseus’ Inspire XC1 clip-on earbuds, which have dual drivers. They even earned a spot on CNET’s best clip-on earbuds list and are probably the best clip-on buds at their price right now.  

Amazon’s Big Spring Sale just kicked off, and it’ll be around through March 31. Right now, all three models are discounted to $100 to $123, bringing them near their all-time low prices. 

That’s a deal I’d highlight on its own, but if you click through to any of those models’ Amazon product pages and look closely, you’ll see that each is eligible for «one free item» with purchase. 

Read more: Best Wireless Earbuds of 2026

You must click the how to claim link first. Then click a button on the left side of the screen (above the stars for average ratings) to switch the view from «qualifying items» to «benefit items» and see the freebie. The items tend to be Baseus’ entry-level headphones or earbuds, but if you don’t like the free item option with a $120 purchase, you can try the options at lower prices.   

You can read my full reviews of the Inspire XH1 headphones here and the Inspire XP1 earbuds here. And here’s my quick take on the Inspire XC1 earbuds:

Like Baseus’ noise-isolating Inspire XP1 earbuds, which I rated highly, the Inspire XC1 have Sound by Bose and a more premium design than earlier Baseus earbuds. The XC1 don’t sound as good as the XP1, they’re decent open earbuds and are equipped with dual drivers (one is a Knowles balanced-armature driver that helps improve treble performance). While they don’t produce as much bass as noise-isolating earbuds like the Inspire XP1, their bass performance is better than I expected. The buds’ sound is pretty full, especially in quieter environments, though they do better with less bass-heavy material. I did notice a bit of distortion at higher volumes with certain tracks that feature harder-driving bass.

While I slightly prefer the design and fit of Bose’s Ultra Open Earbuds, as well as the design of their case, and think the Bose buds sound more natural and a tad better overall, the much more affordable Inspire XC1 fit comfortably and offer top-tier sound for clip-on open earbuds, as well as decent voice-calling performance with good background noise reduction. And they play louder than the Bose, too.

You can grab the Inspire XH1 for $123, the XP1 for $100 and the XC1 for $110, saving you up to 23%. Just remember to claim your free item with your purchase.

Read more: Best Headphones We’ve Tested

For other audio deals happening now, our CNET shopping experts have rounded up headphones, speakers and earbuds deals across a variety of brands and budgets.

NASA is sending a nuclear-powered spacecraft to Mars. Alongside Tuesday’s announcement of its new Ignition program, which features a planned Moon base and a successor to the International Space Station, the agency revealed the SR-1 Freedom, set to launch in 2028 as the first nuclear-powered craft to leave low Earth orbit.

First, SR-1 Freedom will act as a tech demonstration to show that a nuclear-powered spacecraft is a viable option. If it works, it opens the possibility of deeper space exploration by addressing the range limitations imposed by solar power and liquid fuel.

SR-1 Freedom is also responsible for delivering the Skyfall payload to Mars. Skyfall is a team of helicopters that will scour Mars with sensors to find subsurface ice. It’s no secret that Mars harbors water far beneath the surface, but NASA aims to find a large enough pocket of ice near the surface to help sustain human life in future missions. 

It’s not NASA’s first attempt to solve nuclear space travel. The agency has spent $20 billion over more than a dozen failed attempts with only one nuclear reactor to show for it, which is the SNAP-10A that launched into low Earth orbit in 1965. It operated for 43 days before a high-voltage failure shut it down. The SNAP-10a remains in polar orbit to this day. 

What we know about the SR-1 Freedom

Steve Sinacore, NASA Fission Surface Power program director, told reporters in a news conference that NASA would select a launch vehicle from the available stock and that there would be regulatory and inspection proceedings with the Interagency Nuclear Safety Review Board before any selection.

NASA plans to begin developing hardware for the SR-1 Freedom once the design is finalized. This is expected to take roughly 18 months, with assembly beginning in January 2028. Reactor fueling, texting and assembly will continue until the SR-1 Freedom arrives at its launch site in October 2028. 

SR-1 Freedom is targeting a December 2028 launch, as it’s the next available Mars launch window after the one opening in late 2026. The nuclear-electric engine is expected to produce over 20 Kilowatt-electric units and will be integrated with existing spacecraft technology to make the launch timing more realistic. 

The reactor will be powered by high-assay, low-enriched uranium dioxide fuel and will transfer its heat via heat pipes, protected by a boron carbide radiation shield. The heat is converted to power using the Advanced Closed Brayton Cycle Power Conversion System, which then powers the electric propulsion system at the other end of the spacecraft.

Excess heat is handled with a massive heat sink made of composite materials and titanium. The spacecraft’s brain is located between the heat sink and the propulsion system, and will send data back to Earth.

Why nuclear power? 

«Nuclear power in space doesn’t just enhance space exploration, it enables it,» said Sinacore during the press conference. «Through increased energy density, nuclear power will keep lunar bases operating through the 14-day, 354-hour night.»

One of the big problems with deep space exploration and long-term space exploration is power, and NASA hopes nuclear power can solve it. Solar power is a challenge on the moon due to its two-week-long night cycle. According to Sinacore, you would need «football fields of solar panels» to power a long-term base on Mars. 

The next planet out is Jupiter, where solar panel efficiency drops to 4% when compared to Earth. Once you get beyond Jupiter, solar energy is negligible, making it a poor choice for deep space missions.

NASA currently uses liquid propellant for space flight. That doesn’t work for long-term missions and flights due to its mass fraction, which is fancy math that basically says it’s too heavy for long-term spaceflight. The spacecraft wouldn’t be able to move people and cargo efficiently. Sinacore says these are «physics constraints» rather than engineering problems, and that nuclear power solves them. 

What comes after SR-1 Freedom

SR-1 Freedom marks the beginning of many more projects coming over the next couple of decades. Should the SR-1 Freedom prove successful, the next project would be the Lunar Reactor-1, a nuclear reactor that would serve as the power source for NASA’s upcoming moon base. 

Having a nuclear-powered spacecraft and base on the books would open the door for more of both, including a potential human mission to Mars, bigger and more powerful nuclear reactors, and potential commercial participation from companies wanting to get in on the action. 

Smack dab in the middle of this year’s SXSW festival in Austin, Texas, there was a huge dirt hole in the ground, blocks wide, where there used to be a convention center. The festival’s events continued around it in hotels, but the building’s absence was like a lurking symbol. Of chaos, of disruption. Of the world in 2026, dealing with AI and everything else.

I have no idea what the rest of 2026 will bring, but the vibe I felt at a vibe-filled show made me question how AI can work with our lives, our art and our existence. Instead of fighting it, the conference awkwardly embraced it and challenged it. I saw pockets of work all over the place and wondered about it. Conversations. And how to escape it.

Everyone’s trying to handle a world that’s suddenly way too overloaded with AI, generating documents, images, deepfakes and music, injecting assistant agents into our operating systems, even launching entire unleashed and interconnected agent systems all talking to each other on their own social networks. Job-threatening, constantly shifting, training on our data and aiming for our faces. Do we run from it, try to destroy it, or use art to question and challenge it?

SXSW gave me a lot of the latter, in different slices. 

In my panel I was in at SXSW with Meow Wolf’s Vince Kadlubek and Niantic Spatial’s Dennis Hwang about their experiments overlaying tech into art in physical installations, Kadlubek discussed how AI’s infinite slop creative tool becomes uninteresting over time, while intentional art counteracts that. And that’s exactly how I felt moving through intentionally-made experiences that turned my thoughts about AI inside out, all in different ways.

AI seeping into our gaming chats, for better and worse

In a VR headset in a hotel ballroom, I chatted with cartoon fantasy characters in a whimsical game called Fabula Rasa: Dead Man Talking, made by game studio Arvore. I could make any request or beg as much as I wanted from my cage, where I was held prisoner for offending the King and kept dangling over a monster’s mouth for execution. Could I plead my case to them? The cartoonish VR characters responded, but via generative AI improvising off a script from a writing team, using Claude.

The chats were fun, ridiculous. I made myself an irresponsible magician and leaned into improv with the characters who approached me. None of them disappointed, which is a surprise for dialogue that’s somewhat AI-generated. Most interactions felt frazzled and absurd, but it worked for the style and the humor of it all. There was a bit of a delay for responses to kick in, though, standard-issue for a lot of AI conversations. 

This was the best use of AI I saw. But what could it mean for future games, like RPGs? It’s an unsettling thought if you’re a writer…or, exciting. Indie games could end up finding ways to branch out responsive dialogue in ways that still feel custom-written and crafted. I don’t know. 

On the less successful end was Love Bird, an interactive game show experience directed by Cameron Kostopoulos. I was wowed by the initial onboarding, where the «producers» called me on my phone to interview me. The producer was actually an AI chatbot with a surprisingly rapid response time. I convinced the AI to be a participant, and then was led into a room where I spoke via Xbox controller and headset microphone with a PC game on a monitor, where I was competing with others while carnivorous bird-people threatened to eat us. I’m not sure why, exactly. And I don’t know how it all ended, because my chats with the host and participants fell into broken loops that made us have to quit out early. 

Love Bird was fast-paced and responsive, but also too chaotic and weird, even for someone like me who likes weird. It didn’t feel like it was really paying attention to me, and I didn’t feel like I had space to process. Maybe that’s by chaotic design, but after emerging, it just made me want to feel less AI-spammed and have games that didn’t flood me with as much conversation as this one did. I needed a quiet space. My favorite immersive experiences are often the quiet ones, not the chatty ones.

AI as a personal transformational lens

In one room, I stood at a podium and read a portion of New York Mayor Zohran Mamdani’s acceptance speech from November as, before me, video clips of crowds cheering played on a large video monitor, seemingly reacting to me. A few minutes later, I heard my voice delivering more of Mamdani’s speech, AI-generated in my voice, to film clips of inspirational moments of support. I saw my own face layered into the background of some of these clips, too.

The Great Dictator, directed by Gabo Arora, is a museum-style participatory exploration of the power of rhetoric, provocatively named for the Charlie Chaplin satire about Adolf Hitler. The three speeches you can choose from — Mamdani’s, President Ronald Reagan’s on taking down the Berlin Wall, and Malcolm X’s The Ballot or the Bullet speech — are all picked to represent powerful moments in history, and the exhibit is about embodying history and feeling the power of speech and rhetoric in a personal way — and relating to it from a new, personal, and maybe more empathetic angle. The voice AI was generated by ElevenLabs, and the video clips at the end were hand edited, but with AI overlays of my face handled by Runway. What surprised me was how much I ended up being in historical documents. Is this a deepfake? Is it embodiment? Is it both?

Another art experience embedded me into the work: Spectacular, by Jonathan Yeo. Yeo is an artist from London whose portrait work includes King Charles III, President George W. Bush and designer Jony Ive of Apple renown and has played with tech in many of his installations. This gallery at SXSW, replicated from an exhibit that was in Paris before, used Snap Spectacles AR glasses to melt the real portraits with augmented effects and voice narration from Yeo. And, later on, the portraits began overlaying my own face, transformed in art styles that matched Yeo’s using generative AI trained on his work. At the end, I got a printout of my portrait, «signed» by Yeo himself.

I spoke with Yeo in Austin after experiencing his work. He admitted that AI is a provocation here, but that he wants to own the process that AI is trying to take from our own data everywhere. And he’s trying to apply AI and AR in ways that feel intentional and subtle as ways to help play with and bring the art to life, in museums and elsewhere. But again, like with The Great Dictator, I wondered: How much will «permanent» documents of art and history begin to melt over time with AI? What will be kept intact, and who will enforce the line?

AI as broken manipulator

Wearing a pair of Meta Oakley smart glasses, I stood in a room full of objects on shelves as a voice directed me to open a drawer, find a dollar bill there and put it in a shredder filled with bill fragments. I did it. The AI remarked with pleasant surprise at how compliant I was. From there, I «competed» tasks to prove my value as human labor, graded by an AI that saw my actions through the glasses camera and showed my stats on a TV screen, along with a deepfaked dancing version of myself.

Body Proxy, by Tender Claws, applies Meta’s glasses camera feed into its own art AI app on a phone to explore how AI could make us proxies for physical labor. It’s weird and satirical like some of their other VR work (the game Virtual Virtual Reality, among others), but also pushes at a much bigger question: How much is AI breaking us or manipulating us? How much are we willing to be manipulated?

Escape The Internet (Part One), an interactive game I played in a movie theater at the Alamo Drafthouse, turned similar ideas of manipulation into a social experiment. Created by Lucas Rizzotto, another VR/AR provocateur artist, it involved no headsets or glasses. Instead, everyone in the theater used their own phones to connect to a private server that «ran» the game and gave us little personal avatars, feeding us surveys to collect our personal tendencies and then having us play social voting games to see how we’d polarize on decisions like, for instance, who to kill: one person who shared our political views, or five who didn’t?

It’s all absurd and funny and guided by Rizzotto’s in-person guidance at the front of the theater, and along the way, I thought about how social platforms manipulate us with algorithms. Here, in this room together, we’re encouraged to find each other, recognize each other and love each other. The experience has branching paths and can be replayed, and could re-emerge in future conferences and events. But, again, I asked myself: How much of AI is a game that’s playing me, instead of me playing it?

Design for AI is still unfinished (or nonexistent)

In some of the panels I sat in on, and in conversations I had, I got a creeping sense that AI is moving too fast for artists or ethicists — or anyone else, really — to stop and properly process. One panel exploring The Future Design Language of Robots, with Olivia Vagelos of the Design for Feelings Studio, and Savannah Kunovsky, managing director of Ideo’s emerging technology division, tapped into the assumptions we make about robots. I teamed up with someone next to me to try to dream up ideas to break my assumptions and think freshly about what robots could be.

Kunovsky and Vagelos both agreed that designing for AI presents similar challenges right now, particularly because the tech is moving too fast for design to properly attend to it. But sadly, my attempt to record what they said as a quote was sabotaged by my AI-enabled Meta Ray-Ban glasses, which activated as the microphone when I tried recording a voice memo from the panel on my phone, muting the audio completely because of noise cancellation. Wearables are still broken, too.

Another panel, called Generative Ghosts: AI Afterlives and the Future of Memory, led in part by two Google DeepMind researchers, discussed many fascinating angles on how we can responsibly handle archiving our lives via AI as memories in the future, and who controls that ability. The panel had no specific answers but plenty of questions. And, as my own attempt at recording it was also erased by my activated smart glasses, it gave me an additional level of absurd friction which made me wonder: Will these archived memories eventually be lost, too, from big tech companies that sunset services or introduce noncompatible formats, memory-holing the memories? 

AI is threatening, but often not successful in fulfilling its promises (or threats). Self-driving Waymo cars flooded Austin during SXSW, with my Uber app often pushing them on me instead of human drivers. I gave in and took a few for amusement, but they usually took longer to get where I was going. And, one unfortunate evening, my Waymo took a weird roundabout route that ended up dropping me off a half mile from my destination on the wrong side of the highway.

My favorite SXSW memory was making an old-fashioned collage out of magazine clippings with friends at an art gallery over wine, something that involved no tech at all. We worked our magic with intuition, scissors, old magazines and good conversation. Was it perfect? No. But it cost a lot less than generative AI. Which also makes me wonder if all these AI tools being offered to enhance or supplant creativity are necessary, or whether we’ll just rediscover that we had more tools than we realized all along.