Forgot your password?
Close
typodupeerror

Submission + - Newly Discovered PamStealer Isn't Your Typical macOS Malware (arstechnica.com)

Submitted by Anonymous Coward
An anonymous reader writes: Researchers have found a never-before-seen piece of macOS malware that combines a series of clever tradecraft to infect Macs with stealthy, custom-developed credential-stealing code. The malware is delivered in two stages. The first is distributed in a disk image that masquerades as Maccy, a clipboard manager for Macs. It’s compiled as AppleScript that is notable for the way it delivers the second stage. The malware is named PamStealer because the Rust-written infostealer uses the Pluggable Authentication Modules interface built into macOS to validate the target’s login password before sending it to an attacker-controlled server.

[...] PamStealer shows a native password prompt designed to resemble a system authorization request. Text that appears with the prompt says: “Maccy wants to make changes. Enter your password to allow this.” As noted earlier, once a target complies, the malware validates it locally through the PAM API. “This check is done entirely through PAM: there is no call out to dscl, security, osascript or any spawned process to verify the password, as many commodity macOS stealers do," [said Jamf, a security firm for macOS users]. “The result is a quieter routine that keeps only a verified password, and one fewer process chain for defenders to detect on.”

If the validation fails, PamStealer displays the prompts again until it receives the correct one. Once the target enters the correct password, PamStealer displays a message stating that the file is damaged and can’t be installed. This is designed to be a decoy to prevent the target from suspecting anything is amiss. The malware uses tactics to maximize the information it can steal. One tactic is to request the target grant full disk access to the fake Maccy app. It also contains code designed to access ethereum accounts. The various techniques—particularly the Script Editor lure, a self-contained JXA dropper, a Rust-based second stage, and local validation of credentials through PAM are all noteworthy.

Submission + - Sitting For More Than 30 Minutes At a Time Linked To Higher Risk of Cancer Death (theguardian.com)

Submitted by Anonymous Coward
An anonymous reader writes: Researchers who tracked more than 90,000 people over a decade found that sitting or lying down while awake for more than 30 minutes in one period each day was associated with an increased risk of cancer death. The risk increases for every additional hour of continuous inactivity, the findings suggest. However, the researchers also found breaking up periods of sedentary behavior longer than 30 minutes with bursts of physical activity could help reduce the risk. Getting up every half-hour, even for a short walk around the office, could do wonders for your health, they said.

[...] The findings, published in Plos Medicine, focused on the health effects of prolonged sedentary behaviour on a daily basis. [...] The team analyzed data from wearable devices worn by more than 91,000 UK Biobank participants, who were followed for an average of 12 years. The findings suggest prolonged inactivity lasting more than 30 minutes was associated with cancer risks. Each additional hour of prolonged inactivity every day was associated with a 10% increase in risk of cancer death. However, replacing long spells of inactivity with movement appeared to reduce that risk. Substituting one hour of sedentary behaviour each day with light physical activity, such as ironing or washing up, was associated with a 12% lower risk of cancer death.

Replacing 30 minutes of inactivity each day with 30 minutes of moderate physical activity, such as walking at an average pace, was associated with an 8% lower risk. The risk was 22% lower when five minutes of inactivity was replaced with five minutes of vigorous physical activity each day, the study suggested. There were limitations to the research, including the fact that the researchers performed a statistical analysis of an observational study, so could not prove causation.
Businesses

Labor Force Participation Rate Falls To Lowest In 50 years (cnbc.com) 36

Posted by BeauHD from the mass-exodus dept.
The US unemployment rate fell to 4.2% in June largely because 720,000 people left the labor force, pushing participation to 61.5%. Excluding the Covid-era jobs market, that's the lowest participation rate since June 1976. CNBC reports: The decline in the labor force marks a "massive exodus" driven by multiple factors, said Mike Reid, head of U.S. economics at RBC. "The unemployment rate fell to 4.2% as both the number of unemployed workers and the size of the labor force pulled back," Reid wrote in a post-report commentary. "This may well be a story of retirements but could also be a story of prior job seekers dropping out of the labor force."

[...] [T]he rolls of those counted as not in the labor force, a group that includes the unemployed and those not looking for work, jumped by 832,000. And while the establishment survey, which counts jobs filled, showed growth for the month of 57,000, the survey of households, which counts the actual level of those working, tumbled by 507,000. On a year-over-year basis, the labor force is down by just over 1 million, while the level of the employed also has fallen by 1.06 million and the ranks of the unemployed have risen by 40,000. The employment-to-population ratio slipped to 59% in June, the lowest since October 2021. All that has happened while the unemployment rate has risen by just one-tenth of a percentage point to 4.2%.

The drop in participation is sometimes attributed to a shrinking immigrant population and retiring baby boomers and Gen Xers. However, in June the biggest plunge came from what is defined as "prime age" workers, or those between the ages of 25 and 54. That rate fell 0.6 percentage point to 83.3%, its lowest since December 2023. "Looking at the statistics now, that argument doesn't hold up so well," North said of the retirement and immigration rationale. "I hate to use the word 'alarming,'" he added, but said the numbers are cause for concern.
Security

AI Agent Executes 'First' End-To-End Ransomware Attack 17

Posted by BeauHD from the first-of-its-kind dept.
Sysdig says it has documented the first ransomware attack carried out end to end by an AI agent, which autonomously exploited exposed systems, stole credentials, established persistence, compromised a production database, and destroyed data. The research team named the attacker "JadePuffer" and said it gained initial access to an internet-facing Langflow instance by exploiting CVE-2025-3248. "The most striking characteristic, however, was the LLM's behavior," Sysdig director of threat research Michael Clark said in a blog post. An anonymous reader quotes an excerpt from The Register: JadePuffer's "self-narrating" payloads "contained natural language reasoning, target prioritization, and the kind of detailed annotations that human operators don't often write but LLM-generated code produces reflexively," Clark added. "The operation also adapted in real time, retrying failed steps within refined parameters. In one sequence, it went from a failed login to a working fix in 31 seconds." After exploiting CVE-2025-3248, a missing authentication vulnerability in Langflow that allows remote, unauthenticated attackers to execute arbitrary Python on the host, the AI agent began scanning for and collecting secrets, including LLM provider API keys, cloud credentials "with explicit coverage of Chinese providers" including Alibaba, Aliyun, Tencent, and Huawei, while also scanning for AWS, Azure and Google Cloud Platform, cryptocurrency wallets, and database credentials.

The AI also installed a crontab entry on the Langflow server to maintain persistence and call back to the attacker's infrastructure every 30 minutes. JadePuffer's intended target was a separate internet-exposed production server running a MySQL database and an Alibaba Nacos configuration service, we're told. Nacos is an open-source service-discovery and dynamic configuration platform developed by Alibaba and used in the cloud provider's microservices applications. The agent connected to the server's exposed MySQL port using root credentials, although Sysdig doesn't know how the attacker obtained them. These credentials weren't stolen from the victim's environment.

JadePuffer then attacked Nacos via multiple vectors including an authorization bypass flaw (CVE-2021-29441) and forging a valid JSON web token (JWT) using Nacos's default signing key. Additionally, using its root database access, the LLM injected a backdoor administrator into the Nacos backing database. It ultimately encrypted all 1,342 Nacos service configuration items using MySQL's built-in AES encryption function, and created an extortion demand, ransom note, Bitcoin payment address, and a Proton Mail contact [...]. However, according to the threat hunters, the victim can't recover the encrypted data, even if they paid the ransom demand, because the agent escalated "from row-level deletion to dropping entire database schemas, narrating its own targeting rationale," without backing up any of the encrypted data.

Comment Re:We're onboard (Score 1) 79

Just curious since I don't have solar or battery but like the idea... Did you greatly oversize your system? Is your worst case usage significantly higher than your typical usage?

We put 22 panels on our roof, which was probably a few more than we "needed", strictly speaking. We probably would have been fine with 18 or 19 but installation costs are significant and we didn't want to do it piecemeal. Better to do it all at once and be done with it.

As for oversizing, I don't think greatly oversizing your setup makes a lot of sense since most utilities are no longer paying out for over-production, just capping you at the limit of your consumption.

With that said, it never hurts to have some reserve capacity in case you want to add a battery to the system. Or if you buy a kiln.
AI

Godot Game Engine No Longer Accepts AI Code 36

Posted by BeauHD from the suffering-from-slop dept.
The Godot Foundation will stop accepting AI-authored code, agent-submitted pull requests, and AI-generated text in contributor communications after maintainers were overwhelmed by low-effort submissions. "It is time for us to recognize that these problems aren't going away and therefore we need to take steps to reduce the burden on maintainers while ensuring we still have a pipeline to mentor new contributors to become future maintainers," the Godot Foundation said in a blog post. Contributors may still use AI for limited "menial things" if they disclose it, but humans must understand, own, and be able to fix the code they submit. PC Gamer reports: The Foundation says the pileup of Godot pull requests pending review isn't all bad: It's a sign that interest in using and contribution to Godot is increasing. But the influx of contributions authored or submitted by AI is sapping the projects' maintainers of their willingness to confront the "already tedious" work of reviewing pull requests. "If your feedback on PRs is just being absorbed by a machine and not going towards mentoring a potential future maintainer, it becomes much harder to justify spending your free time on PR review," the Foundation said.

As the problem becomes increasingly unsustainable, the Godot Foundation says it's in the process of updating its contribution policies, focusing on "adding barriers to low-effort slop" contributions, encouraging maintainers to review code, developing new contributors into future maintainers, and crucially, requiring that all contributions come from humans who are accountable for their code -- and fixing it if it fails. "AI cannot take responsibility, and we can't trust heavy users of AI to understand their code enough to fix it," the Foundation said.

The Foundation says we can expect Godot's contributing policy to soon include explicit rejections of AI-authored code, noting that contributors should only use AI assistance for "menial things" and must disclose its use. Additionally, the Foundation will reject any AI-generated text in human-to-human communications, saying it's "a basic principle of respect" -- though it says machine translations "are still acceptable" if the original text was human-authored. "Things change every day with respect to the current suite of AI tools available," the Foundation said. "We will continue taking a conservative approach in our policies towards them, but we will re-evaluate as things evolve."

Submission + - AI Agent Executes 'First' End-To-End Ransomware Attack (theregister.com)

Submitted by Anonymous Coward
An anonymous reader writes: They're not bad; they're just prompted that way. 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. The security shop’s research team named the agentic intruder JadePuffer and said it gained initial access to an internet-facing Langflow instance by exploiting CVE-2025-3248, and then ran a fully automated attack. “The most striking characteristic, however, was the LLM's behavior,” Sysdig director of threat research Michael Clark said in a blog about the agentic ransomware and extortion operation.

JadePuffer’s “self-narrating” payloads “contained natural language reasoning, target prioritization, and the kind of detailed annotations that human operators don’t often write but LLM-generated code produces reflexively,” Clark added. “The operation also adapted in real time, retrying failed steps within refined parameters. In one sequence, it went from a failed login to a working fix in 31 seconds.” After exploiting CVE-2025-3248, a missing authentication vulnerability in Langflow that allows remote, unauthenticated attackers to execute arbitrary Python on the host, the AI agent began scanning for and collecting secrets, including LLM provider API keys, cloud credentials “with explicit coverage of Chinese providers” including Alibaba, Aliyun, Tencent, and Huawei, while also scanning for AWS, Azure and Google Cloud Platform, cryptocurrency wallets, and database credentials.

The AI also installed a crontab entry on the Langflow server to maintain persistence and call back to the attacker’s infrastructure every 30 minutes. JadePuffer’s intended target was a separate internet-exposed production server running a MySQL database and an Alibaba Nacos configuration service, we’re told. Nacos is an open-source service-discovery and dynamic configuration platform developed by Alibaba and used in the cloud provider’s microservices applications. The agent connected to the server's exposed MySQL port using root credentials, although Sysdig doesn’t know how the attacker obtained them. These credentials weren’t stolen from the victim’s environment.

JadePuffer then attacked Nacos via multiple vectors including an authorization bypass flaw (CVE-2021-29441) and forging a valid JSON web token (JWT) using Nacos's default signing key. Additionally, using its root database access, the LLM injected a backdoor administrator into the Nacos backing database. It ultimately encrypted all 1,342 Nacos service configuration items using MySQL's built-in AES encryption function, and created an extortion demand, ransom note, Bitcoin payment address, and a Proton Mail contact [...]. However, according to the threat hunters, the victim can’t recover the encrypted data, even if they paid the ransom demand, because the agent escalated “from row-level deletion to dropping entire database schemas, narrating its own targeting rationale,” without backing up any of the encrypted data.

Submission + - AI Agent Executes 'First' End-To-End Ransomware Attack (theregister.com)

Submitted by Anonymous Coward
An anonymous reader writes: They're not bad; they're just prompted that way. 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. The security shop’s research team named the agentic intruder JadePuffer and said it gained initial access to an internet-facing Langflow instance by exploiting CVE-2025-3248, and then ran a fully automated attack. “The most striking characteristic, however, was the LLM's behavior,” Sysdig director of threat research Michael Clark said in a blog about the agentic ransomware and extortion operation.

JadePuffer’s “self-narrating” payloads “contained natural language reasoning, target prioritization, and the kind of detailed annotations that human operators don’t often write but LLM-generated code produces reflexively,” Clark added. “The operation also adapted in real time, retrying failed steps within refined parameters. In one sequence, it went from a failed login to a working fix in 31 seconds.” After exploiting CVE-2025-3248, a missing authentication vulnerability in Langflow that allows remote, unauthenticated attackers to execute arbitrary Python on the host, the AI agent began scanning for and collecting secrets, including LLM provider API keys, cloud credentials “with explicit coverage of Chinese providers” including Alibaba, Aliyun, Tencent, and Huawei, while also scanning for AWS, Azure and Google Cloud Platform, cryptocurrency wallets, and database credentials.

The AI also installed a crontab entry on the Langflow server to maintain persistence and call back to the attacker’s infrastructure every 30 minutes. JadePuffer’s intended target was a separate internet-exposed production server running a MySQL database and an Alibaba Nacos configuration service, we’re told. Nacos is an open-source service-discovery and dynamic configuration platform developed by Alibaba and used in the cloud provider’s microservices applications. The agent connected to the server's exposed MySQL port using root credentials, although Sysdig doesn’t know how the attacker obtained them. These credentials weren’t stolen from the victim’s environment.

JadePuffer then attacked Nacos via multiple vectors including an authorization bypass flaw (CVE-2021-29441) and forging a valid JSON web token (JWT) using Nacos's default signing key. Additionally, using its root database access, the LLM injected a backdoor administrator into the Nacos backing database. It ultimately encrypted all 1,342 Nacos service configuration items using MySQL's built-in AES encryption function, and created an extortion demand, ransom note, Bitcoin payment address, and a Proton Mail contact [...]. However, according to the threat hunters, the victim can’t recover the encrypted data, even if they paid the ransom demand, because the agent escalated “from row-level deletion to dropping entire database schemas, narrating its own targeting rationale,” without backing up any of the encrypted data.
The Almighty Buck

Meta Is Charging a Subscription for Smart Glasses Features (wired.com) 17

Posted by BeauHD from the subscription-fatigue dept.
Meta is introducing a subscription for expanded access to advanced smart-glasses features. According to Wired, "[U]sers will need the Meta One Premium Plan to unlock expanded access to some features for their smart glasses, whether it's the Ray-Ban, Oakley, or Meta-branded version." They'll still be usable with a subscription, but "certain features will be limited," the report says. From the report: Specifically, a feature called Conversation Focus, which boosts the audio of the person you're speaking with so you can hear them better in loud environments. You'll get three hours per month without a subscription, but if you want to use it more often, then you'll need to pay up. Though even then, you're still capped at 15 hours. Subscribing also nets you "Premium Device Support," where you'll get faster access to what Meta says are "human experts" trained on the smart glasses' features, should any problems arise. Guess humans are better at some things after all.

A Meta spokesperson tells WIRED that this is "not an AI rate limit." Rate limits are common on other AI platforms -- users get free access to a feature until they hit a certain cap, then they'll need to subscribe to use it more until the limit resets at the end of the month. However, the Conversation Focus feature runs on-device, meaning it doesn't need to head to Meta's servers for AI processing. There's no real-time way to monitor how many hours you've used Conversation Focus, but you'll receive a notification when you get near the limit.

"The subscription supports that ongoing work and gives power users expanded access along with premium device support," the spokesperson says. "We're going to start testing new optional subscription plans that offer more premium features and advanced capabilities for those who want to unlock more from our apps and AI glasses."
AI

OpenAI 'In Early Talks To Give 5% Stake To US Government' 63

Posted by BeauHD from the financial-stakes dept.
OpenAI is reportedly in early talks to give the U.S. government a 5% stake, potentially alongside similar contributions from other major AI companies. "Such a deal would help improve the industry's relations with the Trump administration and could help garner political support by sharing wealth generated by the AI boom with the public," reports The Guardian. From the report: [OpenAI CEO Sam Altman] and other OpenAI bosses have suggested that each of the biggest AI developers in the US should give 5% to their equity to an investment vehicle such as the Alaska Permanent Fund, a sovereign fund that invests US oil wealth into stocks and pays dividends to the state, the FT reported.

The talks are "conceptual" and in early stages, it said, and any deal could require an act of Congress to implement. Both OpenAI and Anthropic have previously suggested in policy papers that a public or sovereign wealth fund may be required in the future to distribute shares to the public. In April, OpenAI said that a "public wealth fund" could provide "every citizen -- including those not invested in financial markets -- with a stake in AI-driven economic growth." Further reading: Bernie Sanders Unveils $7 Trillion Plan To Give Americans Control of AI Industry

Comment Re:The reason I got it (Score 2) 79

We did more or less the same calculations and decided to install ~10KW of solar panels on our place, grid-tied, no battery yet.

We're producing way more than we're using which should knock our power bill down nearly to zero if not zero.

Power isn't going to get any cheaper and even at the current pricing we're figuring roughly a 10-year payoff.

We'll probably be getting battery rack in the next year so, but we'll see how it goes.

Comment Re:The reason I got it (Score 1) 79

Here in WA state the power companies have to buy excess power exported to the grid at a 1:1 cost.

In other words, if they charge us 20 cents per KWh, they have to pay us (actually credit us) with 20 cents per KWh for every KWh we send them. Credits roll over month-to-month all year long until March when they reset the credits back to zero.

They won't send us a check but our power bill should be at or near zero going forward.

Comment We're onboard (Score 1) 79

We just got done installing almost 10KW worth of solar on our roof and even on cloudy days it produces more than we use- typically way more than what we consume. We've been exporting 25KWh or more every day, even on cloudy days.

The 2nd day it was installed was a nice sunny day and we generated 52KWh while only using about 5 or 6KWh total, so we're expecting our bills will be near or at zero, plus we'll have credits from the exported power to cover the times when we're not producing more than we use.
Privacy

WhatsApp Usernames Are Already Raising Impersonation Red Flags (techcrunch.com) 21

Posted by BeauHD from the PSA dept.
An anonymous reader quotes a report from TechCrunch: WhatsApp this week started rolling out username reservations ahead of the broader launch planned later this year. The feature -- which lets people find and message each other by handle instead of phone number -- is already raising impersonation concerns, drawing scrutiny from security experts and regulators in India, the app's largest market, with more than 500 million users. The rollout marks a shift in how people identify one another on WhatsApp. Instead of relying on phone numbers as the primary identifier, users will increasingly interact through platform-managed usernames, a change that Meta says improves privacy but that critics argue could create new opportunities for impersonation.

[...] Asked about how it protects against impersonation, Meta told TechCrunch it reserves usernames for public figures, government entities, and "some variations" of those names so only the legitimate owner can claim them. The company did not explain, however, how it decides which lookalike usernames get proactively reserved and which don't. The concerns have already reached regulators in India, where cyber fraud schemes frequently exploit messaging platforms to impersonate police, banks, and government officials. [...] Rachel Tobac, chief executive of SocialProof Security, called usernames a net privacy gain because they reduce the need to share phone numbers, which can expose users to SIM-swap attacks, phishing, and account takeovers. Still, she said, lookalike usernames still create opportunities for impersonation. "Ultimately, usernames are a great idea to avoid leaking your phone number to folks you don't know, but it's important to verify identity with the username function too," Tobac told TechCrunch. Her advice for most users: Pick a username that isn't easily guessable, so it's harder for attackers to find you, message you cold, or harass and spam you.

[...] The Mozilla Foundation said the introduction of usernames is likely to bring new tradeoffs. "Increased scams and impersonation from fake handles are potentially a big one," it told TechCrunch. "Checking a phone number can be a useful verification tool, but these harms are also permitted by the platform's fundamental design choices." Mozilla also flagged a broader interoperability question -- one worth logging if you're building on top of, or competing with, Meta's ecosystem. While letting users claim their existing Facebook and Instagram usernames may cut down on impersonation, it also shows how easily Meta can stitch identity together across its own apps, even as users still can't take that identity, or their contacts, to a rival platform. For now, WhatsApp says it is taking a gradual approach to the rollout. "We're taking our time and listening to feedback so that when it rolls out later this year we get it right," the company said in its FAQ.

Submission + - Before Buying Bitcoin Many People Now Ask AI Instead of Google (nerds.xyz)

Submitted by BrianFagioli
BrianFagioli writes: A new study from 5W Public Relations suggests AI may be replacing Google as the first stop for many new cryptocurrency investors. Researchers tested ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews with dozens of prompts about crypto exchanges and found the AI systems frequently made recommendations or issued warnings rather than remaining neutral. Coinbase and Kraken ranked highly while failed companies such as FTX and Celsius continued to be mentioned as cautionary examples years after their collapse. The findings raise a bigger question for the technology industry: are businesses now competing not just for Google rankings, but for AI approval?

Slashdot Top Deals

"The C Programming Language -- A language which combines the flexibility of assembly language with the power of assembly language."

Close