• I bought a Lafufu, a cheap, counterfeit, somewhat creepy version of the viral Labubu toy.
• People around the world are shelling out money to buy Lafufus — it's becoming a retail phenomenon.
• Lafufu lovers told BI they love the toys for how ugly they are, not in spite of it.

As soon as I ripped open the packaging of my brand-new Lafufu, I giggled.

It was my fault. Instead of buying an authentic Labubu from Chinese toymaker Pop Mart, I caved and bought a Lafufu. The sad, budget fuzzy doll stands in for the real deal — a toy that has taken the world by storm and made millions for its Chinese owner.

The term Lafufu is an affectionate nickname for these counterfeit Labubus and is often used in videos on social media platforms like TikTok.

The most obvious knockoff sign was my Lafufu's lopsided face, which made its creepy smile look extremely unsettling. The next thing I spotted was its missing hands, but upon closer inspection, I saw they had been attached backward.

I compared it to my coworker's real Labubu and noticed that the devil really is in the details. The paint job on the legit Labubu was cleaner. My Lafufu, meanwhile, felt flimsier, and its white fur was coarser. The Lafufu's limbs weren't completely mobile — its feet could not be rotated 360 degrees.

And most unfortunately, my poor Lafufu also appeared to have a receding hairline, while the real Labubu had an ample crown of fur on its head.

I snagged the Lafufu toy for 9.77 Singapore dollars, or about $7.50, on local e-commerce platform Shopee. The real toy of this size from Pop Mart costs SG$24.90.

On Shopee, dozens of listings offer various Lafufus, with prices as low as SG$0.60.

Purchasing it was a two-minute affair, a far cry from waiting in line for hours outside a Pop Mart outlet ahead of product drops. In the UK, Pop Mart has temporarily paused physical sales of the toy because queues were getting out of hand.

Labubu and The Monsters toy line was a lucrative product category for Pop Mart in 2024, with sales totaling 3.04 billion yuan, or about $426 million. Pop Mart's stock is up more than 530% in the last year.

While some people may buy Lafufus as a Labubu replacement, others are going out of their way to secure the fake version.

Joey Khong, a trends manager at London-based market research agency Mintel, said, "Like most fakes, Lafufus reflect a combination of systemic market inequalities and genuine human motivations: the desire to belong, to experience joy, and to participate in the cultural moment."

Juda Kanaprach, the cofounder of Singapore-based market research firm Milieu Insight, told me Lafufu is having its "own little cultural moment."

"Whether it's about humor, aesthetics, or just jumping on a trend, everyone's coming at it from a different angle," Kanaprach said. "And that, to me, makes this whole thing more than just a 'fake toy' moment."

Representatives for Pop Mart did not respond to a request for comment from Business Insider.

A cheaper, funnier alternative to Labubu

Lafufu buyers told me they love the fakes because of how ugly they are, not in spite of it.

Miabella Rivera, a college freshman from San Diego, said she got a Lafufu for $12 because the real ones were "impossible to get."

"My Lafufu came without eyes, so I had to superglue them on, but it still turned out really cute," Rivera said.

Renn Lazzerin, who works at an elementary school in Los Angeles as a behavior analyst, has two Lafufus. The first was an unintentional buy — she was scammed by an online seller claiming the doll was authentic.

But soon after, she bought a second Lafufu knowingly, because of how ugly it was.

"The eyes pop out, it has blush blindness, and the teeth are misprinted," Lazzerin said.

"The fake Labubu offers a different serotonin rush than getting an original. It's like, how ugly can it get?" she added.

Khong, the trends manager, said fakes can often look more eccentric or intriguing than real Labubus.

"While anyone with enough money can buy a 'real Labubu' straight from the store, a rare, well-made fake with a unique outfit or expression might require more taste, effort, or insider knowledge to source," he said.

Austin D'Souza, the managing director of Ozzie Collectables, an Australian collectibles store, said Labubus are made with high-quality materials.

"Authentic Labubus are crafted with care and creativity, and counterfeit versions can detract from the unique experience that collectors seek," he said.

"The packaging of genuine products typically features high-quality printing and branding, whereas counterfeit packaging may appear less polished or have inconsistencies," D'Souza added.

Buying a Lafufu has convinced some people to pick up real Labubus

Despite being obvious fakes, Lafufus may be driving up demand for authentic Labubus.

"Even though they're fake, they keep the brand in people's minds. Everyone's sharing them online, making fun of them, and talking about them, which keeps the attention and interest around the real Labubu going," said Kanaprach, from Milieu Insight.

"Buying a fake Labubu was not good enough for me, so I punched the real one to make me feel better about owning just one," said Kimberly Hernandez, a special education assistant from Los Angeles.

The popularity of counterfeits comes with risks

Kanaprach said Pop Mart needs to stay ahead of the game because too many fakes could flood the market. At some point, well-made fakes may become indistinguishable from the real deal.

"Labubu just needs to keep things special, always come up with new designs, keep it limited, and remind people why the original is still worth it," she said.

Khong said a larger problem for brands like Pop Mart is keeping their core fans engaged and satisfied "while riding the volatility of popularity or trend cycles."

While I see the allure of a Labubu — despite never having purchased one myself — I think I'm satisfied with my SG$9.77 Lafufu.

After a few days of staring at its lopsided grin, its deformities have even become endearing.

• Johan Helberg was asleep at home when a 440-foot-long cargo ship crashed into his backyard.
• The NCL Salten was transporting goods to the Norwegian town of Orkanger when it ran aground.
• The ship's owner said it had run aground twice in 2023 and 2024.

A 440-foot-long cargo ship ran aground beside a house in Byneset, Norway, on Thursday morning, local time.

"If the ship had hit the rocky cliff right next to it, it would have lifted up and hit the house hard. It wasn't many meters off," Johan Helberg, the owner of the house, told local newspaper Nidaros.

Helberg said he was asleep when the ship ran aground and did not know what happened until his neighbor alerted him.

"I thought, who in the world rings the doorbell at 5:45 in the morning? I looked out the window, and he said: 'Haven't you seen the ship?'" Helberg told The New York Times in an interview published Thursday.

There were 16 men aboard the NCL Salten, Helberg said in his interview with the Times. The boat was captained by a Norwegian, and its crew comprises Russians and Ukrainians, Helberg said.

Helberg told the Times that his neighbor, Jostein Jørgensen, was "in shock all day" after seeing the ship plow into their backyard.

Jørgensen told the local media outlet TV 2 that he heard the ship at around 5 a.m. local time.

"When I looked out the window, I saw a boat moving at full speed towards shore," Jørgensen said, adding that he expected the ship to turn course initially.

But the ship only stopped moving when it was about "six to eight meters" from Helberg's house wall, Jørgensen told TV 2.

The NCL Salten was transporting goods to Orkanger, a town in Trondheim, Norway, when the crash occurred, per TV 2's report. The ship is owned by Baltnautic, a Lithuanian shipping company.

Baltnautic CEO Bente Hetland told the Times that "nobody was injured in the grounding." She added that the company does not know "what caused the incident and are awaiting the conclusion of the ongoing investigation."

Hetland told TV 2 that the NCL Salten had run aground twice before, both times in Norway. The ship ran aground in Hadsel in 2023 and in Ålesund in 2024.

Baltnautic did not respond to a request for comment from Business Insider.

The Norwegian Coastal Administration said on Thursday that no injuries or oil spills had been reported. It added that Baltnautic and the salvaging company it hired could not "pull the ship off the ground at high tide" with a tugboat on Thursday evening.

"Geotechnical investigations will be carried out, and the shipping company's salvage company is awaiting the results of these to determine whether special considerations need to be taken when the ship is to be pulled off. We expect the investigations to take some time," the statement continued.

• Tom Cruise does his own stunts, and it's remarkable what he's been able to pull off.
• Hanging from the side of a plane, skydiving, climbing the world's tallest building — he's done it all.
• He also jumped off the roof of the Stade de France to mark the end of the 2024 Paris Olympics.

Tom Cruise has been one of the biggest names in Hollywood since the 1980s, and as his star power grew, so did his ambitions. 

He started to do a lot of his own stunts when appearing in action blockbusters like "Top Gun," "Mission: Impossible," and "Minority Report." Now, stunts have become Cruise's calling card.

His ambitiousness also bled into real life at the 2024 Paris Olympics' closing ceremony when he jumped off the roof of the Stade de France.

Now, the actor has added another stunning aerial sequence to his list of feats, in his latest movie: "Mission: Impossible - The Final Reckoning."

Here are the best stunts of Cruise's career, ranked.

12. For the cargo-plane crash in "The Mummy," Cruise did the stunt inside a NASA plane that trains astronauts for zero gravity.

In 2017's "The Mummy," Cruise finds himself stuck in a cargo plane as it crashes. To pull off a scene like this, actors would typically film it in a controlled setting like a sound stage surrounded by a green screen.

Not Cruise, though.

The star shot the scene in a plane that NASA uses to train astronauts.

The scene was filmed in the plane which had to go up to 25,000 feet to get the look that Cruise was in zero gravity. The plane then did a free fall for 22 seconds.

Cruise did the flight four times to pull off the scene.

11. Cruise flew a helicopter in "Mission: Impossible — Fallout."

For the thrilling helicopter-chase scene in the finale of "Fallout," Cruise spent 16 hours a day training to get to the required 2,000 hours to fly a helicopter on his own.

But Cruise didn't just fly the helicopter. He also pulled off a 360-degree corkscrew dive in it, which would challenge even the most veteran pilot.

10. Cruise is really in an F/A-18 jet for the flight scenes in "Top Gun" Maverick" and had to deal with the G-forces.

When you see Cruise and the cast looking like they are battling G-forces in the jets, complete with distorted faces, it's because they really were.

Cruise and the cast went through training so their dogfight scenes could look as realistic as possible — which meant sitting in the F/A-18 jets as they were spun around and took dramatic dives.

9. Cruise jumped off the Stade de France during the closing ceremony of the 2024 Paris Olympics.

The 2024 Paris Olympics closing ceremony went big on the Americana to mark Los Angeles hosting the 2028 games, with Cruise pulling off a stunt straight out of a Hollywood blockbuster.

He jumped off the roof of the Stade de France and descended into France's national stadium in front of thousands in a stunt that blurred the lines between himself and the daring characters he's known for playing. 

When Cruise got to the stage, Los Angeles Mayor Karen Bass and Olympian gymnast Simone Biles gave him the Olympic flag. After it was attached to a motorbike, the actor drove out of the stadium to start the flag's journey to America. He was later shown jumping out of a plane and turning the Hollywood sign into the Olympic rings. 

They're not the most extreme stunts of Cruise's career, but perfectly captures his showmanship.

8. Cruise climbed a 2,000-foot cliff in "Mission: Impossible 2."

In the opening scene of 2000's "M: I 2," Cruise is seen climbing a cliff. And yes, that's really him.

Cruise scaled the cliff in Utah with nothing but a safety rope. He also did a 15-foot jump from one cliff to another.

7. Cruise held his breath for six minutes for an underwater stunt in "Mission: Impossible — Rogue Nation."

In one scene, Cruise's Ethan Hunt has to dive into an underwater safe to retrieve the computer chip that will lead him closer to the villain.

Along with having to hold his breath the whole time, he must keep away from a large crane that's circling around the safe.

For the scene, Cruise first jumped off a 120-foot ledge. Then, in a 20-foot deep-water tank, Cruise held his breath for six minutes.

6. Cruise broke his ankle jumping between buildings while making "Mission: Impossible — Fallout."

Tom Cruise loves to run in his movies; it's become his trademark. But his ability to continue running came into question after a stunt went wrong on the set of "Fallout."

While jumping from one building to another, Cruise hit the wall of the building the wrong way and broke his ankle.

The accident halted production for months and doctors told Cruise his running days might be over. But, six weeks later, Cruise was back on set doing sprints.

5. Cruise climbed the tallest building in the world for "Mission: Impossible — Ghost Protocol."

The Burj Khalifa in Dubai is the tallest building in the world, and Cruise climbed it.

For "Ghost Protocol," the actor's climb got him up to 1,700 feet in the air.

He also fell four stories down by rappelling on the surface of the building.

4. Cruise did 500 skydives and over 13,000 motocross jumps for the thrilling motorcycle stunt in "Mission: Impossible — Dead Reckoning Part 1."

For the latest "M:I" movie, Cruise once again pushed himself.

And one stunt in particular is definitely up there as one of his craziest ideas yet: driving a motorcycle off a cliff.

The star did 500 skydives and over 13,000 motocross jumps to prepare for the stunt. And that wasn't just so Cruise had the skill and comfort to pull off the stunt; the training also made it possible for director Christopher McQuarrie and his crew to map out camera angles to capture it. 

The stunt was then done on the first day of principal photography.

"We know either we will continue with the film or we're not. Let's know day one!" Cruise told "Entertainment Tonight" on why it was done on the first day.

Cruise ended up doing the stunt six times on the day of shooting.

 

3. Cruise hung on the side of a plane as it took off for "Mission: Impossible — Rogue Nation."

Cruise clung to the side of a massive Airbus A400M plane as it took off and went up to 1,000 feet dealing with speeds of 100 knots.

To protect the actor, he was secured with a wire attached to the plane. He also had special contacts on to protect his eyes from debris.

Cruise did this stunt eight times.

2. Cruise navigated between two flying planes in "Mission: Impossible - The Final Reckoning."

In the climax of "The Final Reckoning," Hunt holds on to a biplane during takeoff, before eventually punching its pilot, throwing him out of the vehicle, and climbing into another plane being flown by the villain, Gabriel (Esai Morales).

The ambitious nature of the scene is what we've come to expect from Cruise, but seeing him cling on for dear life above the valleys of South Africa is nothing short of astounding.

Yes, he was strapped to the two different vehicles during multiple takes to achieve the sequence, but watching it unfold on the big screen is still breathtaking.

1. Cruise did 106 skydives with a broken ankle to pull off the HALO jump in "Mission: Impossible — Fallout."

While Cruise was healing the broken ankle he sustained earlier in the "Fallout" production, he went and pulled off the most amazing stunt he's done in his career so far.

In the movie, Cruise's character and CIA tagalong August Walker (Henry Cavill) decide to do a HALO jump — a high-altitude, low-open skydive, in which you open your parachute at a low altitude after free-falling for a period of time — out of a giant C-17 plane to get into Paris undetected.

Cruise did this for real by executing the jump 106 times over two weeks, many of them done during golden hour, a very brief period of perfect lighting that occurs just before sunset.

• UBS is using AI to create avatar videos from analysts' notes.
• 36 analysts covering a range of sectors are taking part in the Swiss bank's initiative.
• UBS told Business Insider that clients have been seeking more video options.

Banks are using AI to save their analysts' time while giving clients what they want.

Bank of America uses "Banker Assist" to aggregate information to offer insights unique to each client, while Goldman Sachs has a "GS AI Assistant" that functions as an in-house ChatGPT for staff.

Swiss bank UBS has gone further, using AI to generate avatars of analysts that explain their research to clients, and it's planning to do this more.

The Swiss bank started using AI avatars of some analysts in January. About 36 UBS analysts, or 5% of its total, have volunteered to take part. They cover sectors including technology, consumer goods, and energy.

UBS's use of AI avatars was first reported by The Financial Times.

Using OpenAI and Synthesia tools, a script is generated in a matter of seconds that is then edited by staff.

Scott Solomon, head of global research technology at UBS, told Business Insider that his team started creating videos of analysts a decade ago, but capacity restrictions meant they were capped at about 1,000 annually.

Analysts were writing an average of two notes a week but would only go to the video studio once a quarter, he said.

The new tools are "enabling somebody to use a capability in video that they weren't really able to use before," he said.

It also gives clients another way to digest information and meet their rising demand for video, Solomon said.

He compared an avatar to other parts of an analyst's toolkit. "When an analyst joins UBS, we give them Excel, we give them our authoring platform, we give them a CRM [customer relationship management] tool so they can talk to clients. I want them to have an avatar," he said.

Solomon said the next step would be integrating the technology so that a video can eventually be created when an analyst publishes a note — without the need for editing.

He said he hoped this would become possible by the end of the year.

Even if the process was fully automated, UBS said analysts will still assess a video based on their notes before it is sent to clients.

Solomon said that ideally, the avatars would eventually become part of the onboarding process, so that whenever a note is published, there's a video too.

The next step would be integrating this capability directly into the authoring platform.

"We have the script generator, we have the ability to send the script to generate the avatar, and then we obviously have the ability to deliver the avatar to clients," Solomon said.

"We want to string all that together so that as they're writing the note, they can get the video with it as well. Our goal is absolutely not to do 50,000 videos a year, but clearly there's an opportunity to do more videos than we are today."

• Systems engineer Michael Floyd worked at Lockheed Martin for four years.
• Floyd was the breadwinner for his family, but chose to become a stay-at-home dad.
• He and his wife budgeted for the transition and even built a chicken coop to counter egg prices.

This as-told-to essay is based on a conversation with Michael Floyd, a 36-year-old stay-at-home dad from Ithaca, New York. It's been edited for length and clarity.

When my wife was pregnant with our child, I was the main breadwinner for our family. We were deciding between two options: putting our daughter in childcare or me becoming a stay-at-home dad.

My wife's plan has always been to become a full-time professor at a top university, so we knew she wouldn't be staying home with our daughter. On the other hand, I was working as a systems engineer for Lockheed Martin, with a six-figure salary.

I took three months of parental leave, and after that, my decision was clear. I left my six-figure engineering job to care for our child while my wife focused on her career. Here's how we decided and budgeted for the change.

New York state childcare would've cost us $30,000 a year

While I was at Lockheed Martin, my wife was working a postdoctoral position at Cornell University. We were also making additional income from our Airbnb rental and two full-time rental properties.

We did calculations, and in our city in New York state, childcare would cost about $30,000 annually.

But I didn't officially decide to be a stay-at-home dad until I spent time caring for my daughter during my parental leave. I noticed just how much attention she needs. I don't trust that even the best childcare worker could attend to my daughter to the degree that I would want.

Plus, until my daughter is of the age where she can raise her voice and let me know something's wrong, it's really hard for me to allow a stranger to watch her. I'm sure 99% of workers are amazing, but I don't want to take a chance on the 1%.

I enjoyed my job, but I left to take care of our daughter full time

I found my job to be fun because it felt like solving Sudoku puzzles all day, but it required me to sit at my desk for 10-hour shifts.

Since I hurt my back during my 6-year service in the military, long periods of sitting or standing make it flare up. After four years doing my job, the last two of which were remote, I felt isolated, so I knew I had made the right decision to leave.

I don't have any plans to return to work at the moment, but it's not off the table for the future.

Since I was the main breadwinner, we had to budget for me to stay home

We canceled our housekeeper and our CrossFit memberships. We also switched from Verizon to T-Mobile, which saves us over $100 each month, and removed one of our three vehicles from our insurance.

I've cut down on our grocery bill by adjusting my diet to rely on more plant-based sources of protein like lentils rather than expensive meats. We also switched to cloth diapers, which we estimate will save us up to $2,000 by the time she's potty-trained.

I even built a chicken coop in response to egg price inflation. I converted the old shed in our backyard into a chicken coop. We currently have five chickens, but our coop can hold up to 20, and we spend $60 on chicken feed monthly.

Being a stay-at-home parent is demanding, but worth it

The biggest challenge of being a stay-at-home dad has been how emotional my daughter's crying rants can be. You'd think that you could just put her in a crib and leave and let her cry it out, but you can't.

It's constant work, but that doesn't affect me much. What affects me is that when she's crying, there's not always something I can do. It's an emotionally difficult experience. Sometimes, it'll be 6 p.m. and she's tired of me and the bottle, and she just wants to be comforted by her mother. In those moments, she's completely inconsolable.

But the best part about being a stay-at-home parent is seeing all of her firsts. When she started smiling after one month, it made everything worth it. There's a lifetime of firsts coming, and I can't wait to see all of them.

If you left a high-paying job to be a stay-at-home parent and would like to share your story, please email the editor, Manseen Logan, at [email protected].

• Coworker.ai nabbed $13 million in VC funding from Triatomic Capital, Abstract Ventures, and Eniac.
• Founded by former Uber managers, the startup builds a general-purpose AI teammate for the workplace.
• Business Insider got an exclusive look at the pitch deck Coworker used to raise its seed round.

Your newest work colleague might not be joining you at the water cooler or at the team's next happy hour: AI agents are infiltrating the workplace, and one startup building those — aptly named Coworker — just cleared a big funding round to put "AI teammates" to work.

Jeff Huber, managing director at Triatomic Capital, led Coworker's $13 million seed funding round, which was announced on Tuesday. Ramtin Naimi of Abstract Ventures, Mallun Yen of Operator Collective, Tim Young of Eniac Ventures, and Clark Golestani, Ken Hausman, and Jack Greenfield of K2 Access Fund also participated in the raise.

Based in San Francisco, Coworker bills itself as a general-purpose AI teammate that can research, plan, and execute high-level work just like an experienced colleague can. The startup said its technology, which about 25 companies have been beta testing since late 2024, has been used across engineering, product management, sales marketing, and operations functions.

For example, a Coworker can act as an engineering teammate to write code, create and review pull requests, and automate release notes, keeping human developers focused on shipping new features. It can also act as a sales teammate, analyzing sales calls and generating proposals and follow-up emails.

Coworker uses its own products at work, which gives the startup a leg up in shipping new features and developing the technology, said CEO and cofounder Alex Calder. For example, Coworker's agents do work like drafting product requirement documents based on customer feedback, creating tickets, writing code, and turning that code into sales talking points, he said.

"In the last six months, we've seen our internal team going from 'AI is good at giving me information' to 'AI is good at using that information to do work for me'," Calder told Business Insider. "That's only possible when you give AI really rich context on your company, your goals, and how you do work."

Calder and his cofounder, Coworker chief product officer Bradford Church, are former Uber managers who led the transportation company's shared rides team.

AI agents are generating strong interest from VCs, and there's high demand, especially for agents that take over rote workplace tasks and free up human employees to focus on more creative and high-impact work. In May, ThriveAI, which builds AI agents that act as junior software engineers, announced a $1.2 million pre-seed round, and AI coding agent startup StackAI announced it landed $16 million.

Another general-purpose AI workplace agent, Artisan, announced in April it raised $25 million.

Calder acknowledged that the AI agent market is getting crowded. When it came to getting VCs excited, focusing on customers helped Coworker close the round, he said.

"There are so many jaw-dropping AI demos these days that VCs are totally desensitized to them," he said. "What they really care about is how customers are actually using the product and whether it can solve real problems at scale. I think what worked in our favor is that we were able to show the real impact Coworker is having inside large companies during our fundraise."

Coworker previously raised $3.5 million in pre-seed funding from Soma Capital, Focal VC, Mischief, and Karman Ventures.

Check out the 14-slide presentation Coworker used to raise $13 million in seed funding.

• Google announced 100 updates at I/O, aiming to dominate the AI landscape.
• Google's AI model Gemini will integrate into Chrome, challenging OpenAI's ChatGPT.
• Google's strategy shows ambition but risks a lack of focus amid competition from OpenAI.

Google made literally 100 announcements at I/O this week, a clear sign that the tech giant intends to dominate every aspect of AI, from its overhaul of Search to its latest AI models and wearables tech.

The event was packed and, at times, felt electrifying. Google showed impressive stats about how its AI has taken off. It had plenty of far-out goals, too, like building a universal AI assistant and extended reality glasses that give directions in real time.

I/O also showcased Google's vulnerabilities. Some releases clearly overlapped, while arch-rival OpenAI upstaged Google on Wednesday with a big announcement of its own.

With the conference now over, here are six main takeaways.

Google wants a 'total overhaul' of Search

The biggest change touted at I/O was AI Mode — what CEO Sundar Pichai called a "total overhaul" of Google's most iconic feature. In AI Mode, users will have a far more conversational Search experience, asking Google questions directly about what they're looking for.

That's a marked change from the traditional experience of going through a long list of links to find the right answer, which feels more clunky than ever in an age of AI chatbots.

At the same time, AI features like these could cannibalize Google Search and threaten the tech giant's main cash cow, Google Ads. Google risks not figuring out how to heavily monetize these AI tools. That being said, it's already testing ads in AI Mode.

Gemini everywhere

Google's AI model family, Gemini, took center stage at I/O. Google announced that it will integrate Gemini into Chrome, allowing users to chat with its latest AI models while they browse. (The feature rolls out to subscribers this summer.) It's a shot across the bow to OpenAI's ChatGPT, which already has a popular Chrome extension.

I/O also announced an array of updates to its Gemini app, which recently passed 400 million monthly active users — an impressive figure, though still behind ChatGPT. With an update called Personal Context, Gemini app users can get tailored responses based on personal data from Google services, like asking its AI to find a long-lost email.

It's all part of a long-term plan to build a universal AI assistant: what Google calls Project Astra. While it's still unfinished, that plan feels more fleshed out now than when Business Insider tested Astra a year ago.

Soaring AI traction

New AI features are undeniably cool, though Google's AI traction garnered some of the biggest reactions at Pichai's keynote speech on Tuesday.

Onstage, Pichai boasted that the number of tokens generated by Google across all its platforms a month had exploded 50 times to over 480 trillion since last year.

The crowd gasped—it was a big moment. Last year's I/O felt like a giant teaser for coming AI features, with plenty of promise but little to show for it. This year felt different.

Sergey Brin goes founder mode

There was no greater manifestation of Google tripling down on AI than cofounder Sergey Brin crashing a fireside chat with DeepMind CEO Demis Hassabis. That was after Brin wandered around a pavilion trying on a pair of Google's XR glasses.

At the chat, Brin said he goes into the office "pretty much every day now" to work on AI. He also said that retired computer scientists should get back to work to take advantage of the current environment.

Brin has been back at Google since 2023 as the search giant races against AI rivals, and it's obvious he's in "founder mode" — something quite rare at a mature company.

Google's smart glasses are here — sort of

Google let BI briefly try on its prototype Android XR glasses, which have Gemini's AI features and allow users to ask questions. While the tech shows promise, it's still early days. Google staffers asked the throngs of I/O attendees lining up for demos not to ask about price, availability, or battery life.

"We just don't know!" they said.

The prototype glasses feel impressively lightweight — almost too much so, to the point that they felt like they might fall off our faces. The display sits only on the right lens and is practically invisible unless viewed at just the right angle under the right light. It's full-color, but it's small and subtle enough that you might miss the display entirely.

We weren't allowed to view Google Maps or Photos in the glasses like Google showed off in its keynote. Instead, we put on the glasses and walked around a room filled with artwork on the walls and travel catalogs on a table that we could ask Gemini questions about.

While Gemini correctly identified the artwork, it couldn't answer a basic travel query when we looked at the travel catalogs: "What is the cheapest flight to New York next month?" And because the display is only on one side, focusing on it made us feel a bit cross-eyed.

The version we saw isn't the final design. It's missing the coming Warby Parker and Gentle Monster flair, though we did see glimmers of something promising here.

Throwing everything against the wall may or may not work

Google's announcements are undeniably impressive, though some of them felt repetitive. It's hard to understand the difference between Search Live and Gemini Live, for example. Both of them involve chatting with your phone about what it sees through its camera.

Google's strategy of launching literally 100 different things at once could work for the company. It could also signal a lack of focus.

BI was at an I/O panel when the news broke that OpenAI was buying former Apple design chief Jony Ive's hardware startup. Seeing OpenAI upstage Google like that felt a little ominous.

The Google panel BI attended was quite dry and technical, with terms like AI-powered "tool calling" mentioned several times. The contrast with OpenAI's buzzy announcement couldn't be clearer. We even saw several attendees check their phones when the news came out.

Google does have massive advantages in scale and distribution, thanks to Android and Chrome.

Still, it's possible that in the long term, something like an AI-native device that ditches Google's ecosystem altogether eventually takes over.

Investors got a taste of that risk last month, when the stock of Google's parent company, Alphabet, briefly tanked after Apple senior vice president Eddy Cue said search volume was shrinking due to AI.

The America that marched for George Floyd five years ago is gone, buried beneath a backlash that has hardened — for now — into a new political and cultural order.

Why it matters: Floyd's murder by a Minneapolis police officer shocked the national conscience. But what looked like historic momentum for racial justice has collapsed — eclipsed by a reactionary movement backed by the full force of the U.S. government.

• Still, activists aren't giving up: They're recharging and refocusing their efforts — shifting from mass protest to defending what remains, and planting the seeds for what's next.
• The fight has moved from the streets to the margins: In courtrooms, classrooms and city councils, a quieter form of resistance is taking shape — often out of the spotlight, but no less determined.

Zoom in: Civil rights groups like the Southern Poverty Law Center, Urban League, and NAACP are investing in long-term infrastructure — working to build durable political power and economic resilience in Black communities.

• Angela Rye, a political strategist and former adviser to the Congressional Black Caucus, is leading a 12-city "State of the People Power" backed by over 200 local and national partners.
• Rev. Jamal Bryant's boycott of Target over its rollback of DEI commitments — which dented the retailer's Q1 sales — shows how economic pressure is becoming a central tool of resistance.
• Selwyn Jones, George Floyd's uncle and co-founder of the Justice 929 Organization, is launching the MYTH app this summer to help record police interactions and alert emergency contacts in real time.

What they're saying: "Progress isn't a straight line. It swings like a pendulum," NAACP president Derrick Johnson told Axios.

• "And for some people, especially younger folks, it can feel like we're going backward. But the truth is we're still perfecting democracy, and the Black community has always been at the center of that work."

Flashback: While the killings of Breonna Taylor and Ahmaud Arbery stirred anger and protests in early 2020, it was Floyd's murder on May 25 — captured on camera and seen around the world — that ignited a global uprising.

• Statues toppled. Streets filled. Cities pledged reforms. Fortune 500 companies embraced diversity initiatives.
• For a moment, it felt like transformative change was coming.

Five years later, the pendulum has swung hard in the opposite direction.

• DEI: On his first day in office, President Trump ordered a government-wide purge of DEI programs and offices — the opening salvo in a systemic effort to dismantle the racial justice agenda that emerged in 2020.
• Civil rights: The Trump administration has moved aggressively to unravel President Lyndon Johnson's civil rights legacy, including by reorienting DOJ priorities to focus on "anti-white racism."
• History: Trump ordered a federal review of Confederate monuments toppled during the 2020 protests, targeting what he called a "concerted and widespread effort to rewrite our Nation's history."
• Police reform: Days before the anniversary of Floyd's murder, the Justice Department scrapped proposed consent decrees for the Minneapolis and Louisville police departments — and dropped nearly a dozen other investigations into alleged police abuse.
• Refugee policy: The administration has effectively ended most refugee programs — except for one tailored to white South African farmers, justified by Trump's false claims of "white genocide."

Zoom out: The racial justice backlash hasn't been confined to government.

• Major corporations that once championed diversity initiatives have slashed DEI staff, removed racial equity language from mission statements, and dropped even the appearance of activism.
• Open racism, antisemitism, and white nationalism have flourished online, with viral incidents — like the cases of Shiloh Hendrix and Karmelo Anthony — fueling toxic tribalism and fundraising.
• Prominent MAGA influencers have even launched a campaign to convince Trump to pardon Derek Chauvin, the police officer convicted of murdering Floyd.

The big picture: Advocates, experts and Floyd family members tell Axios that the 2020 racial reckoning has a mixed legacy, with victories often overlooked amid today's backlash.

• Most Americans say the heightened focus on race and racial inequality following Floyd's death did not lead to improvements for Black Americans, according to a February survey by the Pew Research Center.
• But civil rights attorney Benjamin Crump, who has represented the families of Floyd, Taylor, and countless others in the Black Lives Matter era, argued there has been incremental progress — especially in police accountability.

In the five years since Floyd's death, dozens of cities and states have passed bans on no-knock warrants, expanded crisis response teams and introduced civilian review boards — wins drowned out by public fatigue.

• The NAACP's Johnson acknowledged that fatigue, but he pushed back against the idea that people have stopped fighting.
• "No one is resting," he stressed. "We've earned the right to reflect. But we are still organizing, still fighting — because not only do our lives depend on it, this democracy does too."

Facing a revolt from Miami Republicans, President Trump salvaged his giant spending plan in Congress late Wednesday by ensuring the death of a Chevron oil deal in Venezuela that the lawmakers lividly opposed.

Why it matters: Trump's decision was a matter of political necessity and a nod to his secretary of state, Marco Rubio — a longtime critic of Nicolas Maduro's socialist dictatorship in Venezuela and Chevron's oil export license that helps enrich the regime.

• Trump's move also offered a window into the last-minute dealmaking that saved his priority legislation in the House.
• "Ultimately, he trusts Marco," a senior White House official said of the president.
• "The pro-Maduro Biden oil license in #Venezuela will expire as scheduled next Tuesday May 27th," Rubio announced late Wednesday on X.

The intrigue: The decision marked an abrupt reversal of Trump's special Venezuela envoy, Ric Grenell, who'd announced the day before that the administration would grant a 60-day extension of Chevron's license to export oil from Venezuela.

• As a special envoy, Grenell wants to engage with Maduro.
• As secretary of state, Rubio wants to enforce policies on Venezuela set in Trump's first term.
• Grenell's announcement Tuesday blindsided officials at the White House, the Treasury and State departments, and Rubio's fellow Cuban-Americans from Florida in the House: Carlos Gimenez, Mario Diaz-Balart and Maria Elvira Salazar — all critics of Maduro's regime.

Zoom in: With a razor-thin GOP margin in the House, Speaker Mike Johnson and administration officials knew Wednesday they couldn't lose the three Miami representatives' votes on Trump's big tax-cut and spending bill. Kentucky Rep. Thomas Massie was already a no for other reasons.

• "We just got three new no votes on the 'One Big Beautiful Bill,' " a second White House official groused Wednesday morning. "The Cubans plus Massie kill the bill."
• "Marco was apoplectic," a person who spoke with the secretary of state told Axios.

Zoom out: The Venezuelan and Cuban exile communities share a common bond — relatives who escaped leftist regimes. The Miami Cuban-American Republicans are under pressure at home over the Trump administration's deportation policies and its elimination of immigration protections for thousands of Venezuelans.

• As anti-socialist hardliners, they don't want Chevron to operate in Venezuela and enrich Maduro's regime, which is propped up by Cuba's intelligence services.
• In February, the lawmakers agreed to support Trump's budget plans in return for the president canceling Chevron's license, set to expire Tuesday.
• With that deadline in mind, Grenell negotiated with Caracas, secured the release of an American prisoner, and relayed Trump's interest in extending Chevron's license temporarily. But the timing of Trump's bill gave the Miami representatives leverage against those plans.

Inside the room: "The Cubans didn't have to tell us they were a 'no' again. We just knew it," said a third administration official involved in the discussions. "We knew they wouldn't fold on this."

• So Trump — who spent Wednesday afternoon arm-twisting and cajoling conservative House members to back his massive tax-cut and spending plan — had to engage with the Miami representatives as well.

Late Wednesday afternoon, Rubio arrived at the White House for an event honoring the University of Florida's national championship basketball team (Rubio is a Gator). Afterward, he huddled with Trump in the Oval Office to make his case against the oil deal.

• About 6 p.m., Gimenez — an occasional golfing partner of Trump's —called in by phone.
• Deputy White House Chief of Staff James Blair, a congressional liaison, was a constant presence.
• "Marco spoke to [Trump] about why it's good policy. Blair emphasized the need to keep these members happy to get the bill passed. It was a tag-team effort," a senior White House official said.

People briefed on the discussions told Axios that Rubio, Gimenez and White House officials who met with Trump countered the arguments by Grenell, Chevron and its legion of lobbyists and commentators who have warned that China would benefit from a U.S. withdrawal from the oil deal.

• They noted that China didn't significantly expand in Venezuela when Trump first slapped sanctions on Maduro's regime, which owes China as much as $10 billion.
• The oil market is almost glutted, and its $62-per-barrel price is about break-even for producers. So there's no crisis — and Venezuelan oil is more expensive to refine than others because it's so heavy and sulfurous.
• Finally, they reminded Trump that he'd given his word to the Miami-area lawmakers to end Chevron's deal with Venezuela.
• Gimenez declined to comment. Rubio couldn't be reached for comment.

After Wednesday's meeting, hours went by without word from Trump. The Miami representatives didn't want to push him, but they didn't want to get steamrolled, either.

• "When you negotiate with Trump on something like this, you can't make it look like you're negotiating. You have to apply pressure but not say you're applying pressure. It's delicate," a person involved in the talks said.
• Finally, at 10:57 p.m. Wednesday, Rubio posted his statement on X saying the Chevron lease would still expire Tuesday.

It was a sign to the three Miami lawmakers that Trump would honor his promise.

• Thursday morning, they voted yes on the president's big bill.

Google's newest AI video generator, Veo 3, generates clips that most users online can't seem to distinguish from those made by human filmmakers and actors.

Why it matters: Veo 3 videos shared online are amazing viewers with their realism — and also terrifying them with a sense that real and fake have become hopelessly blurred.

The big picture: Unlike OpenAI's video generator Sora, released more widely last December, Google DeepMind's Veo 3 can include dialogue, soundtracks and sound effects.

• The model excels at following complex prompts and translating detailed descriptions into realistic videos.
• The AI engine abides by real-world physics, offers accurate lip syncing, rarely breaks continuity and generates people with lifelike human features, including five fingers per hand.
• According to examples shared by Google and from users online, the telltale signs of synthetic content are mostly absent.

Case in point: In one viral example posted on X, filmmaker and molecular biologist Hashem Al-Ghaili shows a series of short films of AI-generated actors railing against their AI creators and prompts.

I did more tests with Google's #Veo3. Imagine if AI characters became aware they were living in a simulation! pic.twitter.com/nhbrNQMtqv

— Hashem Al-Ghaili (@HashemGhaili) May 21, 2025

Special effects technology, video-editing apps and camera tech advances have been changing Hollywood for many decades, but artificially generated films pose a novel challenge to human creators.

• In a promo video for Flow, Google's new video tool that includes Veo 3, filmmakers say the AI engine gives them a new sense of freedom with a hint of eerie autonomy.
• "It feels like it's almost building upon itself," filmmaker Dave Clark says.

How it works: Veo 3 was announced at Google I/O on Tuesday and is available now to $249-a-month Google AI Ultra subscribers in the United States.

Between the lines: Google says Veo 3 was "informed by our work with creators and filmmakers," and some creators have embraced new AI tools. But the spread of the videos online is also dismaying many video professionals and lovers of art.

• Some dismiss any AI-generated video as "slop," regardless of its technical proficiency or lifelike qualities — but, as Axios' Ina Fried points out, AI slop is in the eye of the beholder.
• The tool could also be useful for more commercial marketing and media work, AI analyst Ethan Mollick writes.

It's unclear how Google trained Veo 3 and how that might affect the creativity of its outputs.

• 404 Media found that Veo 3 generated the same lame dad joke for several users who prompted it to create a video of a man doing stand-up comedy.
• Likewise, last year, YouTuber Marques Brownlee asked Sora to create a video of a "tech reviewer sitting at a desk." The generated video featured a fake plant that's nearly identical to the shrub Brownlee keeps on his desk for many of his videos — suggesting the tool may have been trained on them.

What we're watching: As hyper-realistic AI-generated videos become even easier to produce, the world hasn't even begun to sort out how to manage authorship, consent, rights and the film industry's future.

• Two Anthropic researchers talked about how to embrace AI to build a successful career.
• Think about what you're doing and how AI can make it better, one researcher said.
• AI is reshaping job markets, affecting sectors like software engineering and consulting.

The secret to building a career in a world where AI is the main character: Lean into it, say two Anthropic researchers.

In an episode of the "Dwarkesh Podcast" released on Thursday, a pair of the researchers behind Claude — Sholto Douglas and Trenton Bricken — shared three strategies for early careers. They suggested thinking big picture, being lazy, and not letting a previous job stop you from working with AI.

Douglas, who works on reinforcement learning, said everyone should imagine what they want to do, now that AI can help.

"If you had 10 engineers at your beck and call, what would you do?" Douglas said. He added, "What problems, and domains suddenly become tractable? That's the world you want to prepare for."

He suggested that people gain technical depth by studying biology, physics, and computer science and that they think hard about what challenges they want to solve.

Bricken, who researches mechanistic interpretability at the AI company, said college students and young professionals should "be lazier" and outsource more to AI.

"You need to critically think about the things you're currently doing, and what an AI could actually be better at doing, and then go and try it," Bricken said.

The researchers' third piece of advice was about not letting "sunk costs" get in the way. Sunk costs are a concept in which people continue to invest more time and resources because so much has already been spent.

"Whatever kind of specialization that you've done, maybe just doesn't matter that much," Bricken said. "My colleagues at Anthropic are excited about AI. They just don't let their previous career be a blocker."

"It's not as if they were in AI forever," he added.

People across industries are talking about how to AI-proof their careers as AI chatbots and agents become more powerful and capable. The technology is displacing jobs in sectors like software engineering, content creation, and consulting.

Top tech leaders have said all professionals need to think about how AI can improve their workflows.

Last month, Uber's CEO, Dara Khosrowshahi, said people must stop perceiving AI as a "tech thing" and see it as a tool for everyone.

"Within Uber, we're a highly technical company — 30,000 employees — and not enough of my employees know how to use AI constructively," Khosrowshahi said, adding that the company is working to change that.

Nvidia's CEO, Jensen Huang, has repeatedly touted the use of AI agents in companies, saying that they will not only change every job but will also secure employment instead of hurting it.

"AIs will recruit other AIs to solve problems. AIs will be in Slack channels with each other, and with humans," Huang said late last year. "So we'll just be one large employee base if you will — some of them are digital and AI, and some of them are biological."