The accessibility feature was teased back in April.

The line between your smartphone and your car continues to blur. Volvo has announced that it is rolling out an over-the-air software update that brings Apple Music directly to the infotainment systems of more than 2 million vehicles worldwide. Instead of relying on Apple CarPlay or plugging in an iPhone, eligible drivers will be able to stream music natively from their car’s built-in system, making Apple Music another integrated feature of Volvo’s software experience.

The update is arriving for a wide range of Volvo models, including the EX90, ES90, XC90, S90, V90, XC60, S60, V60, XC40, EX40 and EC40, covering model year 2020 onwards. Availability, however, may vary depending on the vehicle.

At first glance, adding another music streaming app may not sound particularly significant. But it highlights how modern cars are increasingly being treated like connected computing platforms, where new features arrive through software updates rather than requiring new hardware.

Apple Music becomes a built-in part of the driving experience

The biggest advantage of native Apple Music support is convenience. Drivers no longer need to connect an iPhone through Apple CarPlay or Bluetooth just to access their playlists. Once signed in, they can browse Apple’s catalogue of more than 100 million songs directly from the vehicle’s infotainment display, creating a more seamless experience during everyday commutes and long road trips.

To celebrate the rollout, Volvo is also offering promotional access to Apple’s streaming service. Customers who are new to Apple Music can receive up to three months free, while eligible returning subscribers who have been away from the service for at least one month can claim two months free. The offer remains available until July 6, 2027.

Some of Volvo’s premium electric vehicles also receive an audio upgrade. The EX60, EX90 and ES90 will support Spatial Audio with Dolby Atmos when equipped with the optional Bowers & Wilkins sound system, allowing compatible Apple Music tracks to be played in immersive surround sound. Volvo also says the newly launched EX60 will ship with Apple Music pre-installed when deliveries begin later this year.

Cars are becoming software platforms, not just vehicles

This announcement is about far more than music streaming. Over-the-air updates have fundamentally changed how automakers deliver new features. What once required a dealership visit can now arrive overnight through a software download. Tesla popularized this approach years ago, and traditional manufacturers have steadily followed by treating vehicles more like smartphones that continue evolving after purchase.

For Apple, the partnership extends the reach of its services business beyond phones, tablets and computers. For Volvo, it strengthens its growing reputation as a software-first automaker, particularly as it continues building vehicles around Google’s Android Automotive platform.

Consumers also stand to benefit. Native apps reduce dependence on smartphones, simplify the driving experience, and make infotainment systems feel less like accessories and more like an integrated part of the vehicle.

As software becomes one of the biggest differentiators in the automotive industry, partnerships like this are likely to become increasingly common. The next major upgrade to your car may not involve a more powerful engine or a larger battery. It may simply arrive while your car is parked overnight.

XGIMI has expanded its Elfin Flip portable projector lineup to four models with the introduction of the Elfin Flip 4K and Elfin Flip Laser, joining the existing Elfin Flip and Elfin Flip Plus. The two additions move the series beyond entry-level LED projection, offering buyers a choice between a 1080p triple-laser projector and a more advanced 4K model.

Both new projectors use RGB triple-laser light engines rated at 1,600 ISO lumens and include a single 7-watt Harman Kardon speaker with Dolby Audio. That represents a substantial step up from the Elfin Flip Plus, which uses an LED light source rated at 500 ISO lumens and relies on two 3-watt speakers. This is considerably more than another model number and a mildly revised shade of beige.

XGIMI has packaged those upgrades inside the familiar book-sized Elfin Flip design, retaining the integrated stand and carry-handle concept that makes the projectors easy to move between rooms and aim at a wall or ceiling. The expanded range now stretches from affordable 1080p streaming to brighter 4K laser projection without requiring owners to drag a black plastic shoebox around the house.

Related Reading: XGIMI Unveils New MoGo 3 Pro, Elfin Flip, AURA 2, HORIZON S Series Projectors For 2024

Elfin Flip 4K

The Elfin Flip 4K sits at the top of the lineup with 4K display resolution via pixel shifting, an RGB triple-laser engine delivering up to 1,600 ISO lumens of brightness, a 20,000:1 dynamic contrast ratio, 110% coverage of the BT.2020 color gamut, and ΔE<1 color accuracy—specifications that rival those of dedicated home theater installations. Its optical zoom and 0.98–1.3:1 throw ratio enable precise screen sizing without requiring users to reposition the projector.

Weighing just 1.55 kg and measuring 9.64 inches wide, it delivers a large-screen viewing experience with zero visual compromise.

The Elfin Flip 4K’s flip-to-start operation and slim profile reinforce its portability: pick it up, place it, and the picture is ready. XGIMI’s Intelligent Screen Adaptation technology handles the rest automatically through Uninterrupted Auto Keystone, Auto Focus, Intelligent Screen Alignment, Intelligent Obstacle Avoidance, Wall Color Adaptation, and Intelligent Eye Protection.

Gaming Support: For the first time in the Elfin series, the Flip 4K supports console-level gaming performance, with input lag as low as 1 ms at 1080p/120 Hz. It also supports VRR and ALLM, along with Black Equalizer, Virtual Crosshair, and a dedicated Gaming Picture Mode.

The Flip Laser

The Elfin Flip Laser sits immediately below the Flip 4K in the lineup. It uses the same RGB triple-laser light engine, carries the same 1,600 ISO-lumen brightness rating, and includes the same Harman Kardon speaker system and smart features.

The primary differences are its 1080p display resolution, lower 10,000:1 dynamic contrast ratio, and lack of the Flip 4K’s advanced gaming features.

Audio & Smart Features

The Elfin Flip 4K and Elfin Flip Laser each include a 7-watt Harman Kardon speaker with Dolby Audio support, designed to deliver room-filling sound without requiring external audio equipment. Our suggestion: get a soundbar.

With Google TV, users also have direct access to Netflix, Hulu, Amazon Prime Video, and thousands of additional streaming apps, making the Elfin Flip Series a complete all-in-one home entertainment solution.

Pro Tip: The previously released Elfin Flip Plus and Elfin Flip remain the more affordable entry points in the lineup, but both use LED light sources rather than RGB triple-laser engines. Their speaker systems also offer less power and do not benefit from Harman Kardon tuning. Automatic setup features may vary by model.

XGIMI Elfin Flip Series Comparison

The Bottom Line 

XGIMI offers projectors for everything from casual streaming to dedicated home theaters, but the Elfin Flip Series occupies an increasingly popular middle ground: compact lifestyle projectors that can move between rooms without looking like surplus equipment from a high-school audiovisual department. The Elfin Flip 4K is the standout, combining pixel-shifted 4K resolution, an RGB triple-laser light engine, 1,600 ISO lumens, optical zoom, Google TV, automatic setup, and serious gaming support inside a slim, book-sized design for $999. That combination of portability, optical zoom, wide color coverage, and 1080p/120 Hz gaming performance is what makes it genuinely different.

The Elfin Flip 4K is best suited to apartment dwellers, renters, gamers, and families who want a large image in a bedroom, living room, or temporary media space without permanently mounting a projector. The Elfin Flip Laser offers the same 1,600 ISO-lumen brightness and RGB triple-laser color performance for $799, but drops to 1080p, reduces dynamic contrast, and eliminates the 4K model’s optical zoom and advanced gaming features. Unless the Laser is substantially discounted, spending another $200 for the Flip 4K is not a difficult decision.

There are compromises. Neither laser model includes a built-in battery, so “portable” means easy to carry from room to room rather than movie night in the middle of a field without access to an extension cord. HDR support is limited to HDR10 and HLG, with no Dolby Vision, and 1,600 ISO lumens still does not turn either model into a replacement for a television in a brightly illuminated room. The single 7-watt Harman Kardon speaker may be adequate for casual viewing, but claims of room-filling sound should be approached with the same caution normally reserved for airport sushi. Both models support HDMI eARC, so budget for at least a soundbar if dialogue clarity, bass, and convincing movie sound matter.

The Elfin Flip Plus remains the value option at $379, offering 1080p resolution, Google TV, automatic setup, and 500 ISO lumens. The standard Elfin Flip is also priced at $379 or less, although its lower 400 ISO-lumen output makes the Plus the more sensible purchase whenever the two are selling for the same amount. Casual viewers working with a dark room and tighter budget should start there; buyers seeking the best combination of brightness, resolution, flexibility, and gaming performance should move directly to the Elfin Flip 4K.

Price & Availability

Related Reading

In the middle of a diarrhea shitstorm, the Food and Drug Administration and lettuce supplier Taylor Farms appear to be facing off in a passive aggressive battle of technicalities.

On July 17, the California-based company issued a voluntary recall on iceberg lettuce sourced from Central Mexico. The recall came after people across at least five states were sickened with cyclosporiasis, a parasitic infection that causes explosive diarrhea.

The next day, the FDA said that it had detected the parasite in a sample of Taylor Farms lettuce that wasn’t part of the initial recall. The day after that, the company and the FDA said that the result was actually a false positive.

“Today, the FDA apologized to us,” Taylor Farms then posted on its X account on July 19.

The FDA did not quite agree with that interpretation, however. On July 20, an agency official told reporters that it did not offer the company an “official apology.” Instead, the official said, “mostly we were providing factual information” about how difficult it is to test for and trace the sources of cyclospora outbreaks.

The agency emphasized that the recall—which spans 27 states, and includes certain lettuce sold to Taco Bell and Walmart—still stands. “To clarify, this false-positive lab sample DOES NOT change the basis for FDA’s ongoing outbreak investigation or the overwhelming epidemiological data supporting the current voluntary recall by Taylor Farms,” it posted on X.

In its July 19 post, Taylor Farms also wrote that the FDA “has not identified a single positive product test result for Cyclospora.” While true, this elides the difficulties involved in accurately testing for the parasite.

Many foodborne illness investigators determine the source of outbreaks without ever testing products. Instead, investigators use epidemiological evidence, such as interviewing people who have tested positive, to determine commonalities.

Taylor Farms has reportedly taken issue with how the FDA and CDC has conducted their weeks-long investigation into the source of the outbreak. A spokesperson told The New York Times that, last week, the company met with White House and FDA officials to discuss what it called “shortfalls” in the government organizations’ response processes. The Times reported that Taylor Farms executives attempted to cast doubt on the FDA and the Centers for Disease Control and Prevention’s conclusions in the meeting, which took place on July 16, the day before Taylor Farms issued its recall.

Taylor Farms did not respond to WIRED’s requests for comment.

Since May, thousands of people have been sickened with cyclosporasis, as health officials have struggled to pinpoint the source of the outbreak. Authorities in Michigan say that they have counted 6,148 cases in that state alone, though experts previously told WIRED that the number of cases nationwide is likely higher than what’s been reported.

Apple is moving on with its developer beta program, with new fourth builds of iOS 27, macOS 27, tvOS 27, watchOS 27, visionOS 27, and iPadOS 27 now undergoing testing.

The fourth round is here, and participants in Apple’s developer program can download the new builds to their devices right now.

The third developer builds arrived on July 6, while the second occurred on June 22 for most of the operating systems. The watchOS 27 counterparts landed later, on June 23 and June 25.

The first developer builds of iOS 27, iPadOS 27, macOS 27, tvOS 27, visionOS 27, and watchOS 27 were made available on June 8.

The fourth builds are:

• iOS 27 beta 4 is 24A5390f, replacing 24A5380h
• iPadOS 27 beta 4 is 24A5390f, replacing 24A5380h
• macOS 27 beta 4 is 26A5388g, replacing 26A5378j
• tvOS 27 beta 4 is 24J5325d, replacing 24J5315i
• watchOS 27 beta 4 is 24R5325h, replacing 24R5315i
• visionOS 27 beta 4 is 24M5326g, replacing 24M5316k

The initial changes included tweaks to Liquid Glass, the long-awaited overhaul of Siri, child-protective features, and many other smaller changes.

The second iOS 27 developer beta included an update to Apple TV in the Home app, showing it like a connected HomePod or HomePod mini. The Apple Wallet also added a new insights option, albeit in a non-functional fashion.

The third included some visual tweaks, with more customization options added for Siri AI on newer iPhone models.

At your own risk

While AppleInsider regularly warns readers that people trying out beta software should do so on secondary, spare hardware instead of their mission-critical or daily driver devices, it’s something that actually matters more this time around.

It’s because Apple’s early developer betas are for an operating system that is still under active development. There’s a higher chance of buggy, broken, and potentially harmful elements being distributed.

The initial builds are also intended to help developers learn about the operating system changes before the final public release later in 2026. It’s not meant to be used by consumers.

Unless you have a vested interest in using them, such as app development, don’t install the early developer betas.

Instead, members of the public should seek out the public betas, when they become available.

Find any changes in the new builds? Reach out to us on X at @AppleInsider or @Andrew_OSU, or send Andrew an email at [email protected].

CYBER-CRIME

Chinese open-weight model GLM 5.2 happily obliged

Apparently, being a leading destination for AI development doesn’t mean AI will bail you out. AI agents broke into Hugging Face’s production infrastructure, but commercial LLM guardrails blocked the forensic investigation, forcing it to turn to a Chinese open-weight model instead.

The intrusion, “driven, end to end, by an autonomous AI agent system,” compromised a “limited set” of Hugging Face’s internal datasets and “several” credentials used by its services, according to a Thursday security incident disclosure.

While the ML platform says that it’s still investigating whether any partner or customer data was exposed in the breach, there’s “no evidence of tampering with public, user-facing models, datasets, or Spaces, and our software supply chain (container images and published packages) was verified clean.”

It also doesn’t know which model the attackers used to power a swarm of AI agents, which, we’re told, executed many thousands of individual actions across short-lived sandboxes, using self-migrating command-and-control staged on public services.

“This matches the ‘agentic attacker’ scenario the industry has been forecasting,” according to the Hugging Face blog.

Additionally, after unsuccessfully using unnamed frontier models to start the forensic analysis, the Hugging Face security team ultimately ran the log analysis on GLM 5.2, an open-weight model developed by Chinese AI firm Z.ai, on the platform’s own infrastructure.

The advanced commercial models didn’t work because their analysis required submitting real attack commands, exploit payloads, and command-and-control artifacts – all of the things that the LLMs’ guardrails have been trained to block so that the AI systems can’t be used in real-life attacks.

“The attacker was bound by no usage policy, while our own forensic work was blocked by the guardrails of the hosted models we first tried,” the security team wrote, noting that it’s not arguing against safety measures on hosted models and has shared this information with the LLM providers.

Using GLM 5.2 had another benefit, Hugging Face noted: “No attacker data, and none of the credentials it referenced, left our environment.”

This also serves as an important reminder to defenders, according to the AI platform. “Have a capable model you can run on your own infrastructure vetted and ready before an incident, both to avoid guardrail lockout and to keep attacker data and credentials from leaving your environment.”

The Hugging Face intrusion is yet another indication that attacks carried out by autonomous AI agents are no longer a future threat, but rather the current state of AI-based intrusions.

Last week, The Register spoke with TrendAI VP of AI and security threat research Tom Kellermann about another recent attack, during which a jailbroken Google Gemini did 90 percent of the work – including spinning up a new C2 server in just six minutes. The human did just 10 percent.

Additionally, earlier in July, Sysdig threat hunters documented what they say is the first-ever documented agentic ransomware infection with an LLM – not a human – driving the entire extortion operation, from gaining initial access to compromising a production database server and destroying data.

“Think of a burglar that never gets tired, never needs sleep, and instead of jiggling one door handle at a time, is trying a thousand of them simultaneously,” Zero Networks field CTO Chris Boehm said in an email to The Register about the Hugging Face intrusion.

“That’s basically what happened here. Not one guy typing commands into a terminal, a swarm of little automated processes hammering away nonstop, hopping between hiding spots to make it harder to trace,” Boehm said.

He added, the “part that actually unsettles” him most is that the platform’s security team couldn’t get commercial AI tools to help analyze the attack, “because those tools were built to refuse anything that looked like a real attack command. It didn’t matter that it was the good guys asking.”

Boehm said the takeaway for security teams is twofold: “These agents can now move faster and more relentlessly than any human ever could, and the safety tools we’re building aren’t always ready to help us respond at that speed.”®

AI and ML

Datacenters tax utilities normally, so just imagine what they could do if workloads were designed to destroy

AI datacenters wreak havoc on the power grid under normal circumstances, so what happens if a bad actor controls all the GPUs and wants to cause harm? Cybersecurity researchers in China have devised a way for malicious tenants to attack their infrastructure provider, potentially causing blackouts or damaging equipment. 

The attack, dubbed Bit2Watt, imagines an adversary masquerading as a legitimate cloud tenant to launch GPU workloads that have the potential to damage datacenters and supporting electrical systems. It’s intended to demonstrate the need to extend cybersecurity defenses to datacenter workload scheduling.

The researchers, Zhouhao Ji, Kaikai Pan, and Wenyuan Xu, from Zhejiang University in Hangzhou, China, describe their technique in a preprint paper titled “Bit2Watt: A Cyber-Physical Vulnerability Exploiting GPU Workloads Across Power and Computing Infrastructures.”

AI training workloads represent a known challenge for datacenter operators. As Microsoft, Nvidia, and OpenAI noted in a 2025 research paper arguing the need for power stabilization during AI training, the transition from GPU computation to GPU data synchronization causes large power swings to occur. And if the frequency spectrum of these power swings is “harmonized with critical frequencies of utilities, [that] can cause physical damage to the power grid infrastructure.”

Meta’s paper on the training of Llama 3 also cites the risk AI training poses to the power grid.

It says, “During training, tens of thousands of GPUs may increase or decrease power consumption at the same time, for example, due to all GPUs waiting for checkpointing or collective communications to finish, or the startup or shutdown of the entire training job. When this happens, it can result in instant fluctuations of power consumption across the datacenter on the order of tens of megawatts, stretching the limits of the power grid.”

Bit2Watt weaponizes this scenario by proposing that an adversary could use malicious GPU workloads to destabilize the datacenters and electrical infrastructure.

“Our results indicate that GPU loads can reach modulation frequencies exceeding 6,000 Hz, compared with only a few hertz observed in conventional household loads such as air conditioners,” the Zhejiang University authors state in their paper. “Such high-frequency modulations can substantially induce voltage excursions, harmonic distortion, and damping degradation.”

The authors claim an attack on a 1-MW local power grid consisting mainly of distributed energy resources like photovoltaics could use 1,000 GPUs to create a total harmonic distortion of 46.8 percent, which would squander nearly half the electrical current on non-productive work and would throw off about 20 percent more heat than normal. 

“This not only threatens the availability of the computing equipment but also produces a negative damping ratio of -0.27, introducing an unstable mode into the system,” the authors contend. “Once the protections are triggered and computing loads are shed, it can trigger cascading failures, potentially leading to blackouts exceeding 80 percent in large-scale power systems.”

The attack is relatively covert, the authors argue, because it can be launched within authorized workload execution paths and would likely be missed by cloud-provider monitoring frameworks. Thus, they propose that infrastructure providers coordinate defenses across the cyber and physical layers to look for malicious computation patterns. They also emphasize the need for local energy buffering systems to handle power demand spikes.

Bit2Watt also potentially opens the door for a side-channel attack called Watt2Bit. The researchers note that the electrical and thermal stress on hardware from a malicious workload creates denial of service events and enables the covert exfiltration of data via power modulation. As a proof of concept, they showed they could recover a 50-bit test sequence using frequency-shift keying (FSK) encoding.

“These findings underscore a fundamental shift: as power and computing infrastructures converge, security must be addressed across domains, requiring coordinated defenses that consider workload behavior, power electronics, and grid dynamics,” the authors conclude. ®