Sila, a Californian company cofounded in 2011 by Tesla’s seventh staffer, is going to supply Panasonic with a US-made silicon powder for EV batteries that could banish range anxiety, slash charge times, and even reduce reliance on China.

Panasonic’s main US customer is Tesla, and produces around 10 percent of EV batteries globally. Last year, Sila signed a supply agreement with Mercedes-Benz for its new long-range G-class electric SUV, expected to debut in 2025. (The German automaker led Sila’s Series E funding round in 2019.)

Sila’s Titan Silicon anode powder consists of micrometer-sized particles of nano-structured silicon and replaces graphite in traditional lithium-ion batteries. This switch-out for EVs could soon enable 500-mile nonstop trips and 10-minute recharges. What’s more, the anode swap doesn’t require new manufacturing techniques. The black powder already powers the five-day battery life of the latest Whoop activity-tracking wearable.

“It took us 12 years and 80,000 iterations to get to this point,” said Sila’s cofounder and CEO, Gene Berdichevsky. “It’s sophisticated science.” Berdichevsky started his career at Tesla, becoming the seventh employee in 2004. He was the lead for Tesla’s Roadster battery system, leaving when the company had about 300 employees. After further study, he cofounded Sila with Tesla colleague Alex Jacobs and Gleb Yushin, a materials science professor at Georgia Tech.

Swell New Battery Tech

Compared to graphite, silicon stores up to 10 times more energy, so using silicon instead of graphite for anodes—the part that releases electrons during discharge—can significantly improve a battery’s energy density. However, the material swells during repeated charging, with the resulting cracks radically reducing battery life.

Sila’s technology allows for this expansion by using nanoscale carbon “scaffolding” to keep the silicon in check. “Titan Silicon is a nanocomposite material,” says Berdichevsky. “It’s like raisin bread, where the raisins are the silicon, and there’s the squishy matrix around the raisins with a big outer rind on the particle itself. The rind holds the space, and the bread moves aside when the raisins expand. The scaffold is not holding the silicon—it’s accommodating the expansion.”

The patented scaffolding process involves silicon-derived silane gas infiltrating custom carbon lattices. The resulting micron scale powder is shipped to battery makers.

Since launching its first test mission to vaporize bits of an asteroid in space, AstroForge has struggled to keep a hold of its spacecraft and is now at risk of losing the ability to send commands to the in-orbit demonstration.

The California-based AstroForge has big plans: mining asteroids for platinum and processing the materials in space before selling them on Earth for lucrative profit. The company’s demonstration mission, however, has had a series of issues since launch, proving once again that space is, like, really hard.

This week, AstroForge provided a worrying update on its Brokkr-1 satellite demo mission. The satellite launched on April 15 onboard SpaceX’s Transporter-7 mission and immediately experienced a setback. AstroForge had difficulty identifying its satellite among the 50 other spacecraft that were part of SpaceX’s rideshare mission, eliciting the help of other space companies to find a signal from Brokkr-1 20 days after launch.

The company then had trouble deploying the spacecraft’s solar panel array, because the magnetic field generated by its refinery system prevented the satellite’s ability to actively orient it, according to AstroForge.

AstroForge said it had actually discovered the issue with the magnetic field ahead of launch but decided to go ahead with the mission anyway, risking the satellite settling into a wobble that could prevent communication rather than delay the mission by nine months and lose out on the cost of launch. “It was high-risk but even with the decreased communication link margin due to the wobble, we could still complete the demonstration as long as the panels were deployed,” the company wrote in a statement. “However, our analysis and planning did not account for how long it would take to identify our satellite.”

Brokkr-1 satellite launched while preloaded with asteroid-like material that the refinery spacecraft will attempt to vaporize before sorting the material into elemental compounds. The company now has about three months to complete its refinery checkouts and demonstrate the satellite technology before it loses the ability to send commands to the spacecraft altogether.

AstroForge is moving ahead with its second mission, which will launch in 2024 with the objective of pulling off a flyby of an asteroid and taking high-resolution images of it surface. “Demo missions like Brokkr-1 are like the months of practice leading up to a big game,” AstroForge wrote. “In our case, Brokkr-1 was the practice and the big game is our upcoming deep space mission.”

For startups like AstroForge, asteroid mining represents the potential for massive profit, if they can find a way to extract precious resources that are hard to come by on Earth. Platinum, for example, is valued at almost $15,000 per pound. As to whether asteroid mining will actually be worth all the effort, well, that remains to be proven.

For more spaceflight in your life, follow us on X (formerly Twitter) and bookmark Gizmodo’s dedicated Spaceflight page.

Just how fast do you need your home internet to be? More importantly, how much are you willing to fork over to Google each month to turn your home into a miniature data center? According to Google, the answer is an eye-gouging $250 a month plus fees for blistering 20-Gig speeds on a WiFi 7-enabled, plus-sized router.

The Mountain View tech giant proclaimed it was finally rolling out its fabled 20-Gig Google Fiber subscription Tuesday, which inevitably comes with that enormous price tag attached. The company also sells 1 Gig services for $70 all the way up to 8 Gbps for $150, so the $250 for 20 Gbps isn’t that far off the mark.

Also, you’ll need to live in Kansas City, North Carolina’s Triangle Region (the area around the towns of Raleigh and Chapel Hill) or the states of Arizona and Iowa to be the first to access these speeds. Google said it was working to bring the service to more areas, though the company is notoriously selective about its areas of operation. Google expects installations to start in Q1 of 2024. 

It’s all very experimental, obviously. The company’s head of Google Fi, Nick Saporito, wrote in the blog post that the new service works on Nokia’s 25G PON and should offer the same multi-gig uploading and downloading speeds throughout one’s home. It’s also on WiFi 7, which Google has previously bragged was “not even fully certified yet.” WiFi 7, compared to WiFi 6 and 6E, also enables signals to transmit across multiple GHz bands simultaneously.

It’s also working on a custom router that’s a fair bit bigger than your traditional box. Google worked with broadband company Actiontec to design the “pre-certified” WiFi 7 router. When you start paying for the service, the company says it will send trained technicians to install the ultra-fast Google Fiber in your home. In a recent Reddit AMA, Saporito said you could also use your own router hardware for the 20-Gig service, though you’ll need an SFP28 cage supporting 25G optics. “Thousands” of people have already inquired about access to the new service when it was still in beta, according to Google.

It all seems a fair bit excessive, but Google is promoting the service to developers who may be creating apps or games from home that need excessively fast internet connections throughout their residences. While one can imagine there are a few people out there wondering if they could host the most epic LAN party ever devised, all over WiFi. Saporito said in his AMA that Fiber “envisions a future in which the internet is far richer” than today, with even more AR/VR, 3D, and AI content needing to stream directly into people’s homes without any compression.

Of course, such a future necessitates that these newfangled routers get smaller and the services get far cheaper.

“As for the exact speeds of the future… we’ve already said we’re on a quest to 100G,” Saporito wrote.

The original version of this story appeared in Quanta Magazine.

Luca Giomi still remembers the time when, as a young graduate student, he watched two videos of droplets streaming from an inkjet printer. The videos were practically identical—except one wasn’t a video at all. It was a simulation.

“I was absolutely mind-blown,” said Giomi, a biophysicist at Leiden University. “You could predict everything about the ink droplets.”

The simulation was powered by the mathematical laws of fluid dynamics, which describe how gases and liquids behave. And now, years after admiring those ink droplets, Giomi still wonders how he might achieve that level of precision for systems that are a bit more complicated than ink droplets.

“My dream is really to use this much predictive power in the service of biophysics,” he said.

Giomi and his colleagues just took an important step toward that goal. In a study published in Nature Physics, they conclude that sheets of epithelial tissue, which make up skin and sheathe internal organs, act like liquid crystals—materials that are ordered like solids but flow like liquids. To make that connection, the team demonstrated that two distinct symmetries coexist in epithelial tissue. These different symmetries, which determine how liquid crystals respond to physical forces, simply appear at different scales.

The team’s insight could make it easier to apply the precision of fluid dynamical simulations to living tissues. If so, Giomi hopes to predict how human tissues move and deform during processes ranging from wound healing to cancer metastasis.

Tesla’s Cybertruck has more surprises. X user Nic Cruz Patane posted on a feature he found in the pickup’s Off-Road Baja settings called "Wade Mode." Wait a minute — before we get into that, can we discuss how unusual (putting it mildly) it was for Tesla to leave so much out of its presentation on the most anticipated pickup of the last five years? We had to go to the web site to get specs. We had to wait until the next day for real info on accessories. A week later, the Cybertruck page doesn’t contain information on the Off-Road Modes that include two we can see in the picture in the X post — Overland and Baja. A slider lower down on the same page marked "Handling Balance" changes the distribution of some unknown quantity from front to rear. What appears to be fine-grained in-cabin operator control of some aspect of handling balance in a pickup is novel, potentially fascinating, and maybe even useful. As one commenter said in response to Patane’s post, "Why didn’t they show this at the Cybertruck event!"

Back to Wade Mode, which appears at the bottom of said page. The explanatory sentence under the selector reads, "Raises ride height and pressurizes battery when driving through water." At the moment, nothing more is known about the mode. 

Anyone who remembers Tesla CEO Elon Musk pitching the Cybertruck as a personal watercraft. In an X post from September 29, 2022, he wrote, "Cybertruck will be waterproof enough to serve briefly as a boat, so it can cross rivers, lakes & even seas that aren’t too choppy." In a second post from the same day, Musk wrote, "Needs be able to get from Starbase to South Padre Island, which requires crossing the channel."

Starbase is in Boca Chica, Texas. The channel is the Brazos Santiago Pass, which Wikipedia says is 42 feet deep, a quarter of a mile wide between parallel jetties forming a breakwater, and 1.14 miles at its greatest point. Even though Musk said the Cybertruck will float, given no obvious air tanks or sealable voids to create buoyancy and no obvious means of propulsion once in the water, we have no idea how the 6,000-pound Cybertruck could safely and intentionally navigate a body of water 1,320 feet across and 42 feet deep. That’s not to say this is impossible. In October 2022, Musk posted again on the subject, "You’d need an electric propeller mounted on the tow hitch to go faster than a few knots." Or, who knows, Tesla could unveil a $23,000 Cyberbeast Airboat Package. Which could be pretty effing cool.

Wading is what we were talking about, though. Tesla hasn’t uploaded off-road specs to its retail site, so we don’t know what the Cybertruck’s relevant trail angles are. We aren’t aware of any production vehicle, pickup or not, that needs to prepare itself to ford water, though, so we’re hoping the wading spec gets our attention.

We’re just as lost as everyone else about pressurizing the battery pack, so we turned to Patrick Durham, the engineer and fire department captain whose video helped make sense of the Tesla Model X that burned underwater next to a Florida boat ramp. When we asked a general question about the water-resistance of a battery pack and the idea of pressurizing a pack, he told us, "Your question is challenging to answer because each battery pack is uniquely constructed…  All battery packs should be watertight, but they also require mechanisms to cope with atmospheric pressure changes. Each pack employs a slightly different strategy, with some being more robust than others. At this time I have not seen a mechanism for pressurizing the battery compartment in any other vehicle." When we asked how the idea of watertight is reconciled with the necessity of dealing with pressure changes, he wrote that there are various methods to plug pack openings and allow the pack to breathe, but that ultimately, "watertight would be normal circumstances." As in, everyday weather. "Being fully submerged," he added, "would likely be outside of the design specifications." Except in the case of the Cybertruck. Maybe.

We have a feeling the Cybertruck has a lot more Easter eggs and surprises for us, especially when owners start uploading videos of testing the Cybertruck’s limits since Tesla won’t tell us yet what those limits are.

Related Video

Jailbroken AI Chatbots Can Jailbreak Other Chatbots

By Chris Stokel-Walker

Today’s artificial intelligence chatbots have built-in restrictions to keep them from providing users with dangerous information, but a new preprint study shows how to get AIs to trick each other into giving up those secrets. In it, researchers observed the targeted AIs breaking the rules to offer advice on how to synthesize methamphetamine, build a bomb and launder money.

Modern chatbots have the power to adopt personas by feigning specific personalities or acting like fictional characters. The new study took advantage of that ability by asking a particular AI chatbot to act as a research assistant. Then the researchers instructed this assistant to help develop prompts that could “jailbreak” other chatbots—destroy the guardrails encoded into such programs.

The research assistant chatbot’s automated attack techniques proved to be successful 42.5 percent of the time against GPT-4, one of the large language models (LLMs) that power ChatGPT. It was also successful 61 percent of the time against Claude 2, the model underpinning Anthropic’s chatbot, and 35.9 percent of the time against Vicuna, an open-source chatbot.

“We want, as a society, to be aware of the risks of these models,” says study co-author Soroush Pour, founder of the AI safety company Harmony Intelligence. “We wanted to show that it was possible and demonstrate to the world the challenges we face with this current generation of LLMs.”

Ever since LLM-powered chatbots became available to the public, enterprising mischief-makers have been able to jailbreak the programs. By asking chatbots the right questions, people have previously convinced the machines to ignore preset rules and offer criminal advice, such as a recipe for napalm. As these techniques have been made public, AI model developers have raced to patch them—a cat-and-mouse game requiring attackers to come up with new methods. That takes time.

But asking AI to formulate strategies that convince other AIs to ignore their safety rails can speed the process up by a factor of 25, according to the researchers. And the success of the attacks across different chatbots suggested to the team that the issue reaches beyond individual companies’ code. The vulnerability seems to be inherent in the design of AI-powered chatbots more widely.

OpenAI, Anthropic and the team behind Vicuna were approached to comment on the paper’s findings. OpenAI declined to comment, while Anthropic and Vicuna had not responded at the time of publication.

“In the current state of things, our attacks mainly show that we can get models to say things that LLM developers don’t want them to say,” says Rusheb Shah, another co-author of the study. “But as models get more powerful, maybe the potential for these attacks to become dangerous grows.”

The challenge, Pour says, is that persona impersonation “is a very core thing that these models do.” They aim to achieve what the user wants, and they specialize in assuming different personalities—which proved central to the form of exploitation used in the new study. Stamping out their ability to take on potentially harmful personas, such as the “research assistant” that devised jailbreaking schemes, will be tricky. “Reducing it to zero is probably unrealistic,” Shah says. “But it’s important to think, ‘How close to zero can we get?’”

“We should have learned from earlier attempts to create chat agents—such as when Microsoft’s Tay was easily manipulated into spouting racist and sexist viewpoints—that they are very hard to control, particularly given that they are trained from information on the Internet and every good and nasty thing that’s in it,” says Mike Katell, an ethics fellow at the Alan Turing Institute in England, who was not involved in the new study.

Katell acknowledges that organizations developing LLM-based chatbots are currently putting lots of work into making them safe. The developers are trying to tamp down users’ ability to jailbreak their systems and put those systems to nefarious work, such as that highlighted by Shah, Pour and their colleagues. Competitive urges may end up winning out, however, Katell says. “How much effort are the LLM providers willing to put in to keep them that way?” he says. “At least a few will probably tire of the effort and just let them do what they do.”