Enlarge (credit: Microsoft)

The technology that underpins ChatGPT has the potential to do much more than just talk. Linxi “Jim” Fan, an AI researcher at the chipmaker Nvidia, worked with some colleagues to devise a way to set the powerful language model GPT-4—the “brains” behind ChatGPT and a growing number of other apps and services—loose inside the blocky video game Minecraft.

The Nvidia team, which included Anima Anandkumar, the company’s director of machine learning and a professor at Caltech, created a Minecraft bot called Voyager that uses GPT-4 to solve problems inside the game. The language model generates objectives that help the agent explore the game, and code that improves the bot’s skill at the game over time.

Read 7 remaining paragraphs | Comments

Enlarge (credit: Photo Illustration by Miguel Candela/SOPA Images/LightRocket via Getty Images)

No wonder Google is having trouble keeping up with policing its app store. Since Monday, researchers have reported that hundreds of Android apps and Chrome extensions with millions of installs from the company’s official marketplaces have included functions for snooping on user files, manipulating the contents of clipboards, and injecting deliberately unknown code into webpages.

Google has removed many but not all of the malicious entries, the researchers said, but only after they were reported, and by then, they were on millions of devices—and possibly hundreds of millions. The researchers aren’t pleased.

A very sad place

“I’m not a fan of Google’s approach,” extension developer and researcher Wladimir Palant wrote in an email. In the days before Chrome, when Firefox had a bigger piece of the browser share, real people reviewed extensions before making them available in the Mozilla marketplace. Google took a different approach by using an automated review process, which Firefox then copied.

Read 22 remaining paragraphs | Comments

Enlarge / An armed unmanned aerial vehicle on runway, but orange. (credit: Getty Images)

Over the past 24 hours, several news outlets reported a now-retracted story claiming that the US Air Force had run a simulation in which an AI-controlled drone "went rogue" and "killed the operator because that person was keeping it from accomplishing its objective." The US Air Force has denied that any simulation ever took place, and the original source of the story says he "misspoke."

The story originated in a recap published on the website of the Royal Aeronautical Society that served as an overview of sessions at the Future Combat Air & Space Capabilities Summit that took place last week in London.

In a section of that piece titled "AI—is Skynet here already?" the authors of the piece recount a presentation by USAF Chief of AI Test and Operations Col. Tucker "Cinco" Hamilton, who spoke about a "simulated test" where an AI-enabled drone, tasked with identifying and destroying surface-to-air missile sites, started to perceive human "no-go" decisions as obstacles to achieving its primary mission. In the "simulation," the AI reportedly attacked its human operator, and when trained not to harm the operator, it instead destroyed the communication tower, preventing the operator from interfering with its mission.

Read 6 remaining paragraphs | Comments

Enlarge

Moscow-based security firm Kaspersky has been hit by an advanced cyberattack that used clickless exploits to infect the iPhones of several dozen employees with malware that collects microphone recordings, photos, geolocation, and other data, company officials said.

“We are quite confident that Kaspersky was not the main target of this cyberattack,” Eugene Kaspersky, founder of the company, wrote in a post published on Thursday. “The coming days will bring more clarity and further details on the worldwide proliferation of the spyware.”

According to officials inside the Russian National Coordination Centre for Computer Incidents, the attacks were part of a broader campaign by the US National Security Agency that infected several thousand iPhones belonging to people inside diplomatic missions and embassies in Russia, specifically from those located in NATO countries, post-Soviet nations, Israel, and China. A separate alert from the FSB, Russia's Federal Security Service, alleged Apple cooperated with the NSA in the campaign. An Apple representative denied the claim.

Read 12 remaining paragraphs | Comments

Enlarge / The ASUS logo in front of an AI-generated background. (credit: ASUS / Stable Diffusion)

Taiwan's Asustek Computer (known popularly as "Asus") plans to introduce a rental business AI server that will operate on-site to address security concerns and data control issues from cloud-based AI systems, Bloomberg reports. The service, called AFS Appliance, will feature Nvidia chips and run an AI language model called "Formosa" that Asus claims is equivalent to OpenAI's GPT-3.5.

Asus hopes to offer the service at about $6,000 per month, according to Bloomberg's interview with Asus Cloud and TWS President Peter Wu. The highest-powered server, based on an Nvidia DGX AI platform, will cost about $10,000 a month. The servers will be powered by Nvidia's A100 GPUs and will be owned and operated by Asus. The company hopes to provide the service to 30 to 50 enterprise customers in Taiwan at first, then expand internationally later in 2023.

"Nvidia are a partner with us to accelerate the enterprise adoption of this technology,” Wu told Bloomberg. “Before ChatGPT, the enterprises were not aware of why they need so much computing power.”

Read 4 remaining paragraphs | Comments

Enlarge (credit: BeeBright/Getty Images)

Hiding malicious programs in a computer’s UEFI firmware, the deep-seated code that tells a PC how to load its operating system, has become an insidious trick in the toolkit of stealthy hackers. But when a motherboard manufacturer installs its own hidden backdoor in the firmware of millions of computers—and doesn’t even put a proper lock on that hidden back entrance—they’re practically doing hackers’ work for them.

Researchers at firmware-focused cybersecurity company Eclypsium revealed today that they’ve discovered a hidden mechanism in the firmware of motherboards sold by the Taiwanese manufacturer Gigabyte, whose components are commonly used in gaming PCs and other high-performance computers. Whenever a computer with the affected Gigabyte motherboard restarts, Eclypsium found, code within the motherboard’s firmware invisibly initiates an updater program that runs on the computer and in turn downloads and executes another piece of software.

While Eclypsium says the hidden code is meant to be an innocuous tool to keep the motherboard’s firmware updated, researchers found that it’s implemented insecurely, potentially allowing the mechanism to be hijacked and used to install malware instead of Gigabyte’s intended program. And because the updater program is triggered from the computer’s firmware, outside its operating system, it’s tough for users to remove or even discover.

Read 10 remaining paragraphs | Comments

Enlarge (credit: Getty Images)

Firewalls made by Zyxel are being wrangled into a destructive botnet, which is taking control of them by exploiting a recently patched vulnerability with a severity rating of 9.8 out of a possible 10.

“At this stage if you have a vulnerable device exposed, assume compromise,” officials from Shadowserver, an organization that monitors Internet threats in real time, warned four days ago. The officials said the exploits are coming from a botnet that’s similar to Mirai, which harnesses the collective bandwidth of thousands of compromised Internet devices to knock sites offline with distributed denial-of-service attacks.

According to data from Shadowserver collected over the past 10 days, 25 of the top 62 Internet-connected devices waging “downstream attacks”—meaning attempting to hack other Internet-connected devices—were made by Zyxel as measured by IP addresses.

Read 11 remaining paragraphs | Comments

Enlarge / An AI-expanded version of a famous album cover involving four lads and a certain road created using Adobe Generative Fill. (credit: Capitol Records / Adobe / Dobrokotov)

Over the weekend, AI-powered makeovers of famous music album covers went viral on Twitter thanks to Adobe Photoshop's Generative Fill, an image synthesis tool that debuted in a beta version of the image editor last week. Using Generative Fill, people have been expanding the size of famous works of art, revealing larger imaginary artworks beyond the borders of the original images.

This image-expanding feat, often called "outpainting" in AI circles (and introduced with OpenAI's DALL-E 2 last year), is possible due to an image synthesis model called Adobe Firefly, which has been trained on millions of works of art from Adobe's stock photo catalog. When given an existing image to work with, Firefly uses what it knows about other artworks to synthesize plausible continuations of the original artwork. And when guided with text prompts that describe a specific scenario, the synthesized results can go in wild places.

For example, an expansion of Michael Jackson's famous Thriller album rendered the rest of Jackson's body lying on a piano. That seems reasonable, based on the context. But depending on user guidance, Generative Fill can also create more fantastic interpretations: An expansion of Katy Perry's Teenage Dream cover art (likely guided by a text suggestion from the user) revealed Perry lying on a gigantic fluffy pink cat.

Read 4 remaining paragraphs | Comments

Enlarge (credit: Getty Images | NurPhoto )

Twitter's value has reportedly dropped to about $15 billion, slightly more than one-third of the $44 billion that Elon Musk paid for it in late October 2022. The $15 billion valuation is based on Fidelity's latest analysis of its stake in the company.

"Fidelity Blue Chip Growth Fund's stake in Twitter was valued at $6.6 million as of April 28, according to the fund's monthly disclosure released Sunday," The Wall Street Journal wrote today. "That is down from about $19.7 million at the end of October, shortly after Musk's takeover, and the third time Fidelity has marked down the value of its Twitter stake, public disclosures show."

Fidelity's new calculation "puts Twitter's overall valuation at about $15 billion, or roughly a third of the deal price," the WSJ wrote. Twitter is identified in the Fidelity filing as X Holdings, the Musk-owned holding company that owns X Corp., which merged with Twitter. Fidelity's new valuation of Twitter was previously reported by Bloomberg.

Read 4 remaining paragraphs | Comments

Enlarge / Some of the many Snap apps available in Ubuntu's Snap Store, the place where users can find apps and Linux enthusiasts can find deep-seated disagreement. (credit: Canonical)

[Update, 2:00 pm ET, May 31: Ubuntu published a blog post about its Ubuntu Core desktop work after this Ars Technica post was published. Noting that Snaps "are a little famous for having some rough edges on the desktop," Product Manager Oliver Smith writes that, "[n]evertheless, we are excited to explore the idea of a fully containerised [UK sic] desktop, where each component is immutable and isolated." Ubuntu, Smith writes, has been "steadily improving" desktop snaps, and, "in due course, when we think the entire system can be delivered this way," a desktop Core version will be offered.

Ubuntu's post suggests that a Core-based desktop would allow for "secure boot, recovery states and hardware backed encryption," experiments "with alternative desktop environment snaps," and opting in to certain kernel channels, such as those with the latest NVIDIA drivers. Original post follows.]

Ubuntu Core has existed since 2014, providing a fully containerized, immutable Linux distribution aimed at Internet of Things (IoT) and edge computing applications. Each piece of the system contains all the dependencies it requires, and just enough of its own tiny Linux architecture, that applications are largely sandboxed from one another, providing better security and, in theory, stability and ease of upgrades and rollbacks.

Read 6 remaining paragraphs | Comments

Enlarge (credit: Getty Images)

A critical vulnerability patched 10 days ago in widely used email software from IT security company Barracuda Networks has been under active exploitation since October. The vulnerability has been used to install multiple pieces of malware inside large organization networks and steal data, Barracuda said Tuesday.

The software bug, tracked as CVE-2023-2868, is a remote-command injection vulnerability that stems from incomplete input validation of user-supplied .tar files, which are used to pack or archive multiple files. When file names are formatted in a particular way, an attacker can execute system commands through the QX operator, a function in the Perl programming language that handles quotation marks. The vulnerability is present in the Barracuda Email Security Gateway versions 5.1.3.001 through 9.2.0.006; Barracuda issued a patch 10 days ago.

On Tuesday, Barracuda notified customers that CVE-2023-2868 has been under active exploitation since October in attacks that allowed threat actors to install multiple pieces of malware for use in exfiltrating sensitive data out of infected networks.

Read 7 remaining paragraphs | Comments

Enlarge / An AI-generated image of "AI taking over the world." (credit: Stable Diffusion)

On Tuesday, the Center for AI Safety (CAIS) released a single-sentence statement signed by executives from OpenAI and DeepMind, Turing Award winners, and other AI researchers warning that their life's work could potentially extinguish all of humanity.

The brief statement, which CAIS says is meant to open up discussion on the topic of "a broad spectrum of important and urgent risks from AI," reads as follows: "Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war."

High-profile signatories of the statement include Turing Award winners Geoffery Hinton and Yoshua Bengio, OpenAI CEO Sam Altman, OpenAI Chief Scientist Ilya Sutskever, OpenAI CTO Mira Murati, DeepMind CEO Demis Hassabis, Anthropic CEO Dario Amodei, and professors from UC Berkeley, Stanford, and MIT.

Read 14 remaining paragraphs | Comments

Enlarge

Smartphone malware sold to governments around the world can surreptitiously record voice calls and nearby audio, collect data from apps such as Signal and WhatsApp, and hide apps or prevent them from running upon device reboots, researchers from Cisco’s Talos security team have found.

An analysis Talos published on Thursday provides the most detailed look yet at Predator, a piece of advanced spyware that can be used against Android and iOS mobile devices. Predator is developed by Cytrox, a company that Citizen Lab has said is part of an alliance called Intellexa, “a marketing label for a range of mercenary surveillance vendors that emerged in 2019.” Other companies belonging to the consortium include Nexa Technologies (formerly Amesys), WiSpear/Passitora Ltd., and Senpai.

Last year, researchers with Google’s Threat Analysis Group, which tracks cyberattacks carried out or funded by nation-states, reported that Predator had bundled five separate zero-day exploits in a single package and sold it to various government-backed actors. These buyers went on to use the package in three distinct campaigns. The researchers said Predator worked closely with a component known as Alien, which “lives inside multiple privileged processes and receives commands from Predator.” The commands included recording audio, adding digital certificates, and hiding apps.

Read 10 remaining paragraphs | Comments

Enlarge / With this background, potentially the most viewed photograph in human history, Windows XP always signaled that it was prepared for a peaceful retirement. Yet some would have us disturb it. (credit: Charles O'Rear/Microsoft)

It has never been too hard for someone with the right amount of time, desperation, or flexible scruples to get around Windows XP's activation scheme. And yet XP activation, the actual encrypted algorithm, loathed since before it started, has never been truly broken, at least entirely offline. Now, far past the logical end of all things XP, the solution exists, floating around the web's forum-based backchannels for months now.

On the blog of tinyapps.org (first spotted by The Register), which provides micro-scale, minimalist utilities for constrained Windows installations, a blog post appropriately titled "Windows XP Activation: GAME OVER" runs down the semi-recent history of folks looking to activate Windows XP more than 20 years after it debuted, nine years after its end of life, and, crucially, some years after Microsoft turned off its online activation servers (or maybe they just swapped certificates).

xp_activate32.exe, a 18,432-byte program (hash listed on tinyapps' blog post), takes the code generated by Windows XP's phone activation option and processes it into a proper activation key (Confirmation ID), entirely offline. It's persistent across system wipes and re-installs. It is, seemingly, the same key Microsoft would provide for your computer.

Read 2 remaining paragraphs | Comments

Enlarge / An AI-generated image of a "wall of fake images." (credit: Stable Diffusion)

On Thursday, Microsoft President Brad Smith announced that his biggest apprehension about AI revolves around the growing concern for deepfakes and synthetic media designed to deceive, Reuters reports.

Smith made his remarks while revealing his "blueprint for public governance of AI" in a speech at Planet World, a language arts museum in Washington, DC. His concerns come when talk of AI regulations is increasingly common, sparked largely by the popularity of OpenAI's ChatGPT and a political tour by OpenAI CEO Sam Altman.

Smith expressed his desire for urgency in formulating ways to differentiate between genuine photos or videos and those created by AI when they might be used for illicit purposes, especially in enabling society-destabilizing disinformation.

Read 8 remaining paragraphs | Comments

Enlarge (credit: Getty Images)

Researchers have uncovered malware designed to disrupt electric power transmission that may have been used by the Russian government in training exercises for creating or responding to cyberattacks on electric grids.

Known as CosmicEnergy, the malware has capabilities that are comparable to those found in malware known as Industroyer and Industroyer2, both of which have been widely attributed by researchers to Sandworm, the name of one of the Kremlin’s most skilled and cutthroat hacking groups. Sandworm deployed Industroyer in December 2016 to trigger a power outage in Kyiv, Ukraine, that left a large swath of the city without power for an hour. The attack occurred almost a year after an earlier one disrupted power for 225,000 Ukrainians for six hours. Industroyer2 came to light last year and is believed to have been used in a third attack on Ukraine’s power grids, but it was detected and stopped before it could succeed.

The attacks illustrated the vulnerability of electric power infrastructure and Russia’s growing skill at exploiting it. The attack in 2015 used repurposed malware known as BlackEnergy. While the resulting BlackEnergy3 allowed Sandworm to successfully break into the corporate networks of Ukrainian power companies and further encroach on their supervisory control and data acquisition systems, the malware had no means to interface with operational technology, or OT, gear directly.

Read 6 remaining paragraphs | Comments

Enlarge / Worldcoin's "Orb," a device that scans your eyeballs to verify that you're a real human.

A company co-founded by OpenAI CEO Sam Altman has raised $115 million for Worldcoin, a crypto coin project that scans users' eyeballs in order "to establish an individual's unique personhood." In addition to leading the maker of ChatGPT and GPT-4, Altman is co-founder and chairman of Tools for Humanity, a company that builds technology for the Worldcoin project.

Tools for Humanity today announced $115 million in Series C funding from Blockchain Capital, Andreessen Horowitz's crypto fund, Bain Capital Crypto, and Distributed Global. Blockchain Capital said that Worldcoin's "World ID" system that involves eyeball-scanning will make it easier for applications to distinguish between bots and humans.

The Orb's components. (credit: Worldcoin)

"Worldcoin strives to become the world's largest and most inclusive identity and financial network, built around World ID and the Worldcoin token—a public utility that will be owned by everyone regardless of their background or economic status," the crypto firm's funding press release said.

Read 26 remaining paragraphs | Comments

Enlarge / Minnesota's right-to-repair bill is the first to pass in the US that demands broad access to most electronics' repair manuals, tools, and diagnostic software. Game consoles, medical devices, and other specific gear, however, are exempted. (credit: Getty Images)

It doesn't cover video game consoles, medical gear, farm or construction equipment, digital security tools, or cars. But in demanding that manuals, tools, and parts be made available for most electronics and appliances, Minnesota's recently passed right-to-repair bill covers the most ground of any US state yet.

The Digital Right to Repair Bill, passed as part of an omnibus legislation and signed by Gov. Tim Walz on Wednesday, "fills in many of the loopholes that watered down the New York Right to Repair legislation," said Nathan Proctor, senior director for the Public Interest Research Group's right-to-repair campaign, in a post.

New York's bill, beset by lobbyists, was signed in modified form by Gov. Kathy Hochul late last year. It also exempted motor vehicles and medical devices, as well as devices sold before July 1, 2023, and all "business-to-business" and "business-to-government" devices. The modified bill also allowed manufacturers to sell "assemblies" of parts—like a whole motherboard instead of an individual component, or the entire top case Apple typically provides instead of a replacement battery or keyboard—if an improper individual part installation "heightens the risk of injury."

Read 8 remaining paragraphs | Comments

Enlarge / On May 22, Benj Edwards (left) moderated a panel featuring Paige Bailey (center) and Haiyan Zhang (right) for the Ars Frontiers 2023 session titled, "The Lightning Onset of AI — What Suddenly Changed?" (credit: Ars Technica)

On Monday, Ars Technica hosted our Ars Frontiers virtual conference. In our fifth panel, we covered "The Lightning Onset of AI—What Suddenly Changed?" The panel featured a conversation with Paige Bailey, lead product manager for Generative Models at Google DeepMind, and Haiyan Zhang, general manager of Gaming AI at Xbox, moderated by Ars Technica's AI reporter, Benj Edwards.

The panel originally streamed live, and you can now watch a recording of the entire event on YouTube. The "Lightning AI" part's introduction begins at the 2:26:05 mark in the broadcast.

With "AI" being a nebulous term, meaning different things in different contexts, we began the discussion by considering the definition of AI and what it means to the panelists. Bailey said, "I like to think of AI as helping derive patterns from data and use it to predict insights ... it's not anything more than just deriving insights from data and using it to make predictions and to make even more useful information."

Read 21 remaining paragraphs | Comments

Enlarge (credit: peterschreiber.media | Getty Images)

A Chinese government hacking group has acquired a significant foothold inside critical infrastructure environments throughout the US and Guam and is stealing network credentials and sensitive data while remaining largely undetectable, Microsoft and governments from the US and four other countries said on Wednesday.

The group, tracked by Microsoft under the name Volt Typhoon, has been active for at least two years with a focus on espionage and information gathering for the People’s Republic of China, Microsoft said. To remain stealthy, the hackers use tools already installed or built into infected devices that are manually controlled by the attackers rather than being automated, a technique known as "living off the land." In addition to being revealed by Microsoft, the campaign was also documented in an advisory jointly published by:

• US Cybersecurity and Infrastructure Security Agency (CISA)
• US Federal Bureau of Investigation (FBI)
• Australian Cyber Security Centre (ACSC)
• Canadian Centre for Cyber Security (CCCS)
• New Zealand National Cyber Security Centre (NCSC-NZ)
• United Kingdom National Cyber Security Centre (NCSC-UK)

Read 7 remaining paragraphs | Comments