OpenAI said a group with apparent ties to China tried to carry out a phishing attack on its employees, reigniting concerns that bad actors in Beijing want to steal sensitive information from top US artificial intelligence companies. From a report: The AI startup said Wednesday that a suspected China-based group called SweetSpecter posed as a user of OpenAI's chatbot ChatGPT earlier this year and sent customer support emails to staff. The emails included malware attachments that, if opened, would have allowed SweetSpecter to take screenshots and exfiltrate data, OpenAI said, but the attempt was unsuccessful.

"OpenAI's security team contacted employees who were believed to have been targeted in this spear phishing campaign and found that existing security controls prevented the emails from ever reaching their corporate emails," OpenAI said. The disclosure highlights the potential cybersecurity risks for leading AI companies as the US and China are locked in a high-stakes battle for artificial intelligence supremacy. In March, for example, a former Google engineer was charged with stealing AI trade secrets for a Chinese firm.

An anonymous reader quotes a report from TechCrunch: U.S. money transfer giant MoneyGram has confirmed that hackers stole its customers' personal information and transaction data during a cyberattack last month. The company said in a statement Monday that an unauthorized third party "accessed and acquired" customer data during the cyberattack on September 20. The cyberattack -- the nature of which remains unknown -- sparked a week-long outage that resulted in the company's website and app falling offline. MoneyGram says it serves over 50 million people in more than 200 countries and territories each year.

The stolen customer data includes names, phone numbers, postal and email addresses, dates of birth, and national identification numbers. The data also includes a "limited number" of Social Security numbers and government identification documents, such as driver's licenses and other documents that contain personal information, like utility bills and bank account numbers. MoneyGram said the types of stolen data will vary by individual. MoneyGram said that the stolen data also included transaction information, such as dates and amounts of transactions, and, "for a limited number of consumers, criminal investigation information (such as fraud)."

TechCrunch's Paul Sawers reports: Fledgling fintech startup OpenBB is revealing the next step in its plans to take on the heavyweights of the investment research world. The company is launching a new, free version of a product that will open its arsenal of data and financial tooling to more users. OpenBB is the handiwork of software engineer Didier Lopes, who launched the Python-based platform back in 2021 as a way for amateur investors and enthusiasts to do investment research using different datasets for free, via a command line interface (CLI). The company went on to raise $8.5 million in seed funding from OSS Capital and angel investors such as Ram Shriram, an early backer of Google. While the community-based, open source project has amassed some 50,000 users, OpenBB has also been building an enterprise incarnation called Terminal Pro. This paid version gives teams access to an interface, pre-built database integrations, an Excel add-in, and various security and support bolt-ons that would appeal to larger businesses. [...]

The all-new OpenBB Terminal -- not to be confused with the previous CLI-based OpenBB Terminal that the startup sunsetted in March -- is a full-fledged web app, though it strips out many of the premium features of Terminal Pro. It's fully customizable, can run on any operating system or platform, and provides access to an AI-enabled OpenBB copilot. Like the previous OpenBB Terminal, the all-new web app is also free to use. OpenBB Terminal is perhaps something of a middle ground between the CLI centricity of the open source project and the bells-and-whistles feature set of the enterprise product.

The OpenBB Terminal serves as a single end point for accessing financial information from some 100 data sources, spanning equity, options, forex, the macro economy, and more. Users can also throw all their new data into the mix -- the community has previously contributed financial datasets such as historical currency exchange rates and crypto pricing data. There are also a slew of extensions and toolkits to bring more functionality to OpenBB -- such as an AI stock analysis agent. Users are free to incorporate their own AI systems and large language models (LLMs), which might be particularly important for security and compliance use cases. But with the OpenBB Copilot, categorized as a "compound AI system," users can run natural-language queries about their data out of the box. While OpenBB has been likened to an "open-source Bloomberg," TechCrunch notes that it's not a direct competitor due to Bloomberg's massive data resources and built-in chat functionality. OpenBB, however, offers flexibility with its open-source platform and customization options.

OpenBB filed for a trademark, but Bloomberg has requested an extension to potentially oppose it, despite the company asserting there's no link between OpenBB and Bloomberg's abbreviation "BBG". Lopes says the name originates from BlackBerry stock, where the founders had lost money during the meme stock craze.

Apple has rolled out an update to macOS 15 Sequoia that addresses compatibility issues with third-party security software that emerged in the initial release. The update, macOS 15.0.1, aims to resolve problems affecting products from CrowdStrike and Microsoft. The compatibility problems had disrupted the functionality of several cybersecurity tools when macOS 15 first launched in September.

U.S. public utility giant American Water says it has disconnected some of its systems after discovering that hackers breached its internal networks last week. From a report: American Water, which supplies drinking water and wastewater services to more than 14 million people across the United States, confirmed the security incident in an 8-K regulatory filing with the U.S. Securities and Exchange Commission on Monday. The New Jersey-based company said in its filing that its water and wastewater facilities are "at this time" not affected and continue to operate without interruption, though the company noted that it's currently "unable to predict the full impact of this incident." American Water said it also notified law enforcement of the intrusion.

The company said it discovered "unauthorized activity" within its networks on October 3 and promptly moved to disconnect affected systems. In a statement on its website, American Water said it is "pausing billing until further notice." "In an effort to protect our customers' data and to prevent any further harm to our environment, we disconnected or deactivated certain systems," Ruben E. Rodriguez, a spokesperson for American Water, told TechCrunch in a statement. "There will be no late charges for customers while these systems are unavailable." Rodriguez declined to state which systems were unavailable and also declined to comment on the nature of the cybersecurity incident.

An anonymous Slashdot reader shared the EFF's "Deeplink" blog post: EFF, along with the ACLU and the ACLU of Mississippi, filed an amicus brief on Thursday asking a federal appellate court to continue to block Mississippi's HB 1126 — a bill that imposes age verification mandates on social media services across the internet. Our friend-of-the-court brief, filed in the U.S. Court of Appeals for the Fifth Circuit, argues that HB 1126 is "an extraordinary censorship law that violates all internet users' First Amendment rights to speak and to access protected speech" online.

HB 1126 forces social media sites to verify the age of every user and requires minors to get explicit parental consent before accessing online spaces. It also pressures them to monitor and censor content on broad, vaguely defined topics — many of which involve constitutionally protected speech. These sweeping provisions create significant barriers to the free and open internet and "force adults and minors alike to sacrifice anonymity, privacy, and security to engage in protected online expression." A federal district court already prevented HB 1126 from going into effect, ruling that it likely violated the First Amendment.

At the heart of our opposition to HB 1126 is its dangerous impact on young people's free expression. Minors enjoy the same First Amendment right as adults to access and engage in protected speech online. "No legal authority permits lawmakers to burden adults' access to political, religious, educational, and artistic speech with restrictive age-verification regimes out of a concern for what minors might see" [argues the brief]. "Nor is there any legal authority that permits lawmakers to block minors categorically from engaging in protected expression on general purpose internet sites like those regulated by HB 1126..."
"The law requires all users to verify their age before accessing social media, which could entirely block access for the millions of U.S. adults who lack government-issued ID..." And it also asks another question. "Would you want everything you do online to be linked to your government-issued ID?"

And the blog post makes one more argument. "in an era where data breaches and identity theft are alarmingly common." So the bill "puts every user's personal data at risk... No one — neither minors nor adults — should have to sacrifice their privacy or anonymity in order to exercise their free speech rights online."

An anonymous reader shared this report from Engadget: Three new theft protection features that Google announced earlier this year have reportedly started rolling out on Android. The tools — Theft Detection Lock, Offline Device Lock and Remote Lock — are aimed at giving users a way to quickly lock down their devices if they've been swiped, so thieves can't access any sensitive information. Android reporter Mishaal Rahman shared on social media that the first two tools had popped up on a Xiaomi 14T Pro, and said some Pixel users have started seeing Remote Lock.

Theft Detection Lock is triggered by the literal act of snatching. The company said in May that the feature "uses Google AI to sense if someone snatches your phone from your hand and tries to run, bike or drive away." In such a scenario, it'll lock the phone's screen.
The Android reporter summarized the other two locking features in a post on Reddit:

• Remote Lock "lets you remotely lock your phone using just your phone number in case you can't sign into Find My Device using your Google account password."

• Offline Device Lock "automatically locks your screen if a thief tries to keep your phone disconnected from the Internet for an extended period of time."

"All three features entered beta in August, starting in Brazil. Google told me the final versions of these features would more widely roll out this year, and it seems the features have begun expanding."

"Pig Butchering Alert: Fraudulent Trading App targeted iOS and Android users."

That's the title of a new report released this week by cybersecurity company Group-IB revealing the official Apple App Store and Google Play store offered apps that were actually one part of a larger fraud campaign. "To complete the scam, the victim is asked to fund their account... After a few seemingly successful trades, the victim is persuaded to invest more and more money. The account balance appears to grow rapidly. However, when the victim attempts to withdraw funds, they are unable to do so."

Forbes reports: Group-IB determined that the frauds would begin with a period of social engineering reconnaissance and entrapment, during which the trust of the potential victim was gained through either a dating app, social media app or even a cold call. The attackers spent weeks on each target. Only when this "fattening up" process had reached a certain point would the fraudsters make their next move: recommending they download the trading app from the official App Store concerned.

When it comes to the iOS app, which is the one that the report focussed on, Group-IB researchers said that the app remained on the App Store for several weeks before being removed, at which point the fraudsters switched to phishing websites to distribute both iOS and Android apps. The use of official app stores, albeit only fleetingly as Apple and Google removed the fake apps in due course, bestowed a sense of authenticity to the operation as people put trust in both the Apple and Google ecosystems to protect them from potentially dangerous apps.
"The use of web-based applications further conceals the malicious activity," according to the researchers, "and makes detection more difficult." [A]fter the download is complete, the application cannot be launched immediately. The victim is then instructed by the cybercriminals to manually trust the Enterprise developer profile. Once this step is completed, the fraudulent application becomes operational... Once a user registers with the fraudulent application, they are tricked into completing several steps. First, they are asked to upload identification documents, such as an ID card or passport. Next, the user is asked to provide personal information, followed by job-related details...

The first discovered application, distributed through the Apple App Store, functions as a downloader, merely retrieving and displaying a web-app URL. In contrast, the second application, downloaded from phishing websites, already contains the web-app within its assets. We believe this approach was deliberate, since the first app was available in the official store, and the cybercriminals likely sought to minimise the risk of detection. As previously noted, the app posed as a tool for mathematical formulas, and including personal trading accounts within an iOS app would have raised immediate suspicion.
The app (which only runs on mobile phones) first launches a fake activity with formulas and graphics, according to the researchers. "We assume that this condition must bypass Apple's checks before being published to the store. As we can see, this simple trick allows cybercriminals to upload their fraudulent application to the Apple Store." They argue their research "reinforces the need for continued review of app store submissions to prevent such scams from reaching unsuspecting victims". But it also highlights "the importance of vigilance and end-user education, even when dealing with seemingly trustworthy apps..."

"Our investigation began with an analysis of Android applications at the request of our client. The client reported that a user had been tricked into installing the application as part of a stock investment scam. During our research, we uncovered a list of similar fraudulent applications, one of which was available on the Google Play Store. These apps were designed to display stock-related news and articles, giving them a false sense of legitimacy."

The Washington Post interviewed Lebanese officials, people close to Hezbollah, and Israeli, Arab and U.S. security officials and politicians about a years-long plan (originated at Mossad headquarters) that ultimately killed or maimed "as many as 3,000 Hezbollah officers and members — most of them rear-echelon figures... along with an unknown number of civilians... when Israel's Mossad intelligence service triggered the devices remotely on September 17." In the initial sales pitch to Hezbollah two years ago, the new line of Apollo pagers seemed precisely suited to the needs of a militia group with a sprawling network of fighters and a hard-earned reputation for paranoia... Best of all, there was no risk that the pagers could ever be tracked by Israel's intelligence services. Hezbollah's leaders were so impressed they bought 5,000 of them and began handing them out to mid-level fighters and support personnel in February. None of the users suspected they were wearing an ingeniously crafted Israeli bomb...

Israeli officials had watched with increasing anxiety as the Lebanese group added new weapons to an arsenal already capable of striking Israeli cities with tens of thousands of precision-guided missiles. Mossad, the Israeli intelligence service responsible for combating foreign threats to the Jewish state, had worked for years to penetrate the group with electronic monitoring and human informants. Over time, Hezbollah leaders learned to worry about the group's vulnerability to Israeli surveillance and hacking, fearing that even ordinary cellphones could be turned into Israeli-controlled eavesdropping and tracking devices. Thus was born the idea of creating a kind of communications Trojan horse, the officials said. Hezbollah was looking for hack-proof electronic networks for relaying messages, and Mossad came up with a pair of ruses that would lead the militia group to purchase devices that seemed perfect for the job — equipment that Mossad designed and had assembled in Israel.

The first part of the plan, booby-trapped walkie-talkies, began being inserted into Lebanon by Mossad nearly a decade ago, in 2015. The mobile two-way radios contained oversized battery packs, a hidden explosive and a transmission system that gave Israel complete access to Hezbollah communications. For nine years, the Israelis contented themselves with eavesdropping on Hezbollah, the officials said, while reserving the option to turn the walkie-talkies into bombs in a future crisis. But then came a new opportunity and a glitzy new product: a small pager equipped with a powerful explosive. In an irony that would not become clear for many months, Hezbollah would end up indirectly paying the Israelis for the tiny bombs that would kill or wound many of its operatives.

Because Hezbollah leaders were alert to possible sabotage, the pagers could not originate in Israel, the United States or any other Israeli ally. So, in 2023, the group began receiving solicitations for the bulk purchase of Taiwanese-branded Apollo pagers, a well-recognized trademark and product line with a worldwide distribution and no discernible links to Israeli or Jewish interests. The Taiwanese company had no knowledge of the plan, officials said... The marketing official had no knowledge of the operation and was unaware that the pagers were physically assembled in Israel under Mossad oversight, officials said... In a feat of engineering, the bomb component was so carefully hidden as to be virtually undetectable, even if the device was taken apart, the officials said. Israeli officials believe that Hezbollah did disassemble some of the pagers and may have even X-rayed them.
"Thousands of Apollo-branded pagers rang or vibrated at once, all across Lebanon and Syria," according to the article, with a short sentence in Arabic that said "You received an encrypted message." The two-button de-encryption procedure "ensured most users would be holding the pager with both hands when it detonated," according to the article, although "Less than a minute later, thousands of other pagers exploded by remote command, regardless of whether the user ever touched his device. The following day, on September 18, hundreds of walkie-talkies blew up in the same way, killing and maiming users and bystanders..."

"As Hezbollah reeled, Israel struck again, pounding the group's headquarters, arsenals and logistic centers with 2,000-pound bombs," the article concludes. And the strike "convinced the country's political leaders that Hezbollah could be put on the ropes, susceptible to a systematic dismantling using airstrikes and, eventually a ground invasion..."

"A cyberattack tied to the Chinese government penetrated the networks of a swath of U.S. broadband providers," reports the Wall Street Journal, "potentially accessing information from systems the federal government uses for court-authorized network wiretapping requests.

"For months or longer, the hackers might have held access to network infrastructure used to cooperate with lawful U.S. requests for communications data, according to people familiar with the matter, which amounts to a major national security risk." The attackers also had access to other tranches of more generic internet traffic, they said. Verizon Communications, AT&T and Lumen Technologies are among the companies whose networks were breached by the recently discovered intrusion, the people said.

The widespread compromise is considered a potentially catastrophic security breach and was carried out by a sophisticated Chinese hacking group dubbed Salt Typhoon. It appeared to be geared toward intelligence collection, the people said... The surveillance systems believed to be at issue are used to cooperate with requests for domestic information related to criminal and national security investigations. Under federal law, telecommunications and broadband companies must allow authorities to intercept electronic information pursuant to a court order. It couldn't be determined if systems that support foreign intelligence surveillance were also vulnerable in the breach...

The hackers appear to have engaged in a vast collection of internet traffic from internet service providers that count businesses large and small, and millions of Americans, as their customers. Additionally, there are indications that the hacking campaign targeted a small number of service providers outside the U.S., the people said. A person familiar with the attack said the U.S. government considered the intrusions to be historically significant and worrisome... "It will take time to unravel how bad this is, but in the meantime it's the most significant in a long string of wake-up calls that show how the PRC has stepped up their cyber game," said Brandon Wales, former executive director at the Cybersecurity and Infrastructure Security Agency and now a vice president at SentinelOne, referring to the People's Republic of China. "If companies and governments weren't taking this seriously before, they absolutely need to now."
Three weeks ago TechCrunch also reported that the FBI "took control of a botnet made up of hundreds of thousands of internet-connected devices, such as cameras, video recorders, storage devices, and routers, which was run by a Chinese government hacking group, FBI director Christopher Wray and U.S. government agencies revealed Wednesday.

T-Mobile experienced three major data breaches in 2021, 2022, and 2023, according to CSO Online, "which impacted millions of its customers."

After a series of investigations by America's Federal Communications Commission, T-Mobile agreed in court to a number of settlement conditions, including moving toward a "modern zero-trust architecture," designating a Chief Information Security Office, implementing phishing-resistant multifactor authentication, and adopting data minimization, data inventory, and data disposal processes designed to limit its collection and retention of customer information.

Slashdot reader itwbennett writes: According to a consent decree published on Monday by the U.S. Federal Communications Commission, T-Mobile must pay a $15.75 million penalty and invest an equal amount "to strengthen its cybersecurity program, and develop and implement a compliance plan to protect consumers against similar data breaches in the future."

"Implementing these practices will require significant — and long overdue — investments. To do so at T-Mobile's scale will likely require expenditures an order of magnitude greater than the civil penalty here,' the consent decree said.
The article points out that order of magnitude greater than $15.75 million would be $157.5 million...

Last week the Register warned "If you're running the Unix printing system CUPS, with cups-browsed present and enabled, you may be vulnerable to attacks that could lead to your computer being commandeered over the network or internet." (Although the CEO of cybersecurity platform watchTowr told them "the vulnerability impacts less than a single-digit percentage of all deployed internet-facing Linux systems.")

But Tuesday generic (Slashdot reader #14,144) shared this new warning from Akamai: Akamai researchers have confirmed a new attack vector using CUPS that could be leveraged to stage distributed denial-of-service (DDoS) attacks. Research shows that, to begin the attack, the attacking system only needs to send a single packet to a vulnerable and exposed CUPS service with internet connectivity.

The Akamai Security Intelligence and Response Team (SIRT) found that more than 198,000 devices are vulnerable to this attack vector and are accessible on the public internet; roughly 34% of those could be used for DDoS abuse (58,000+). Of the 58,000+ vulnerable devices, hundreds exhibited an "infinite loop" of requests.

The limited resources required to initiate a successful attack highlights the danger: It would take an attacker mere seconds to co-opt every vulnerable CUPS service currently exposed on the internet and cost the attacker less than a single US cent on modern hyperscaler platforms.

An anonymous reader quotes a report from The Register: Comcast says data on 237,703 of its customers was in fact stolen in a cyberattack on a debt collector it was using, contrary to previous assurances it was given that it was unaffected by that intrusion. That collections agency, Financial Business and Consumer Solutions aka FBCS, was compromised in February, and according to a filing with Maine's attorney general, the firm informed the US cable giant about the unauthorized access in March. At the time, FBCS told the internet'n'telly provider that no Comcast customer information was affected. However, that changed in July, when the collections outfit got in touch again to say that, actually, the Comcast subscriber data it held had been pilfered.

Among the data types stolen were names, addresses, Social Security numbers, dates of birth, and the Comcast account numbers and ID numbers used internally at FBCS. The data pertains to those registered as customers at "around 2021." Comcast stopped using FBCS for debt collection services in 2020. Comcast made it clear its own systems, including those of its broadband unit Xfinity, were not broken into, unlike that time in 2023. FBCS earlier said more than 4 million people had their records accessed during that February break-in. As far as we're aware, the agency hasn't said publicly exactly how that network intrusion went down. Now Comcast is informing subscribers that their info was taken in that security breach, and in doing so seems to be the first to say the intrusion was a ransomware attack. [...]

FBCS's official statement only attributes the attack to an "unauthorized actor." It does not mention ransomware, nor many other technical details aside from the data types involved in the theft. No ransomware group we're aware of has ever claimed responsibility for the raid on FBCS. When we asked Comcast about the ransomware, it simply referred us back to the customer notification letter. The cableco used that notification to send another small middle finger FBCS's way, slyly revealing that the agency's financial situation prevents it from offering the usual identity and credit monitoring protection for those affected, so Comcast is having to foot the bill itself.

On Wednesday, the SEC filed (PDF) to appeal a 2023 court ruling that determined XRP is not considered a security when sold to retail investors on exchanges. The announcement sent the price of XRP tumbling more than 8%. "XRP, which was created by the founders of Ripple, is the native token of the open source XRP Ledger, which Ripple uses in its cross-border payments business," notes CNBC. "It is the fifth-largest coin by market cap, excluding stablecoins Tether (USDT) and USD Coin (USDC)." CNBC reports: Ripple, the largest holder of XRP coins, scored a partial victory last summer after a three-year battle with the SEC. U.S. District Judge Analisa Torres handed down the decision, which was hailed as a landmark win for the crypto industry. Still, while XRP isn't considered a security when sold to retail investors on exchanges, it is considered an unregistered security offering if sold to institutional investors.

Ripple declined to comment but referred to Wednesday evening posts on X by CEO Brad Garlinghouse and chief legal officer Stuart Alderoty. Alderoty said the company is evaluating whether to file a cross appeal, and called the SEC's decision to appeal "disappointing, but not surprising." The SEC, under Chair Gary Gensler, has become notorious for its refusal to provide clear guidance for crypto businesses, instead opting to regulate by enforcement actions. "XRP's status as a non-security is the law of the land today - and that does not change even in the face of this misguided - and infuriating - appeal," Garlinghouse said on X.

Microsoft has released Windows 11 Dev build 26120.1930, which contains the ability to remap the Copilot key. The changes are rolling out gradually to Dev Insiders with the "Get the latest features as soon as they are available" toggle on. Neowin reports: [H]ere are the updates that are also gradually rolling out, but this time for all Dev Insiders: "We are adding the ability to configure the Copilot key. You can choose to have the Copilot key launch an app that is MSIX packaged and signed, thus indicating the app meets security and privacy requirements to keep customers safe. The key will continue to launch Copilot on devices that have the Copilot app installed until a customer selects a different experience. This setting can be found via Settings - Personalization - Text input. If the keyboard connected to your PC does not have a Copilot key, adjusting this setting will not do anything. We are planning further refinements to this experience in a future flight." Other changes introduced in the build include a new simplified Chinese font, Windows Sandbox improvements, and several bug fixes. Full release notes are available here.

A sophisticated malware strain has infected thousands of Linux systems since 2021, exploiting over 20,000 common misconfigurations and a critical Apache RocketMQ vulnerability, researchers at Aqua Security reported. Dubbed Perfctl, the malware employs advanced stealth techniques, including rootkit installation and process name mimicry, to evade detection. It persists through system reboots by modifying login scripts and copying itself to multiple disk locations. Perfctl hijacks systems for cryptocurrency mining and proxy services, while also serving as a backdoor for additional malware. Despite some antivirus detection, the malware's ability to restart after removal has frustrated system administrators.

Cybersecurity firm IronNet, founded by former NSA director Keith Alexander, has collapsed after failing to deliver on its promise to revolutionize cyber defense. The company, which went public in 2021 with a $3 billion valuation, shut down in September 2023 after running out of money.

IronNet's downfall has left investors and former employees bitter, with some accusing the company of misleading them about its financial health. "I'm honestly ashamed that I was ever an executive at that company," said Mark Berly, a former IronNet vice president. He said the company's top leaders cultivated a culture of deceit "just like Theranos." Critics point to questionable business practices, subpar products, and associations that potentially exposed the firm to Russian influence. The company's board included high-profile national security figures, which helped attract investments and contracts. However, IronNet struggled to secure major deals and meet revenue projections.

Apple just fixed a duo of security bugs in iOS 18.0.1 and iPadOS 18.0.1, one of which might cause users' saved passwords to be read aloud. It's hardly an ideal situation for the visually impaired. From a report: For those who rely on the accessibility features baked into their iGadgets, namely Apple's VoiceOver screen reader, now is a good time to apply the latest update. In typical Apple fashion, the company hasn't released much in the way of details about the first security issue, tracked as CVE-2024-44204, which makes it tougher to understand the conditions under which this vulnerability could be triggered, or how to avoid it until the update is applied. What we do know is that it was characterized as a logic issue, which Apple rectified by improving validation. The disclosure of the bug comes less than a month after iOS 18 and iPadOS 18 debuted. Ironically, this release included Apple's first native password manager, the Passwords app.

An anonymous reader quotes a report from KrebsOnSecurity: Organizations that get relieved of credentials to their cloud environments can quickly find themselves part of a disturbing new trend: Cybercriminals using stolen cloud credentials to operate and resell sexualized AI-powered chat services. Researchers say these illicit chat bots, which use custom jailbreaks to bypass content filtering, often veer into darker role-playing scenarios, including child sexual exploitation and rape. Researchers at security firm Permiso Security say attacks against generative artificial intelligence (AI) infrastructure like Bedrock from Amazon Web Services (AWS) have increased markedly over the last six months, particularly when someone in the organization accidentally exposes their cloud credentials or key online, such as in a code repository like GitHub.

Investigating the abuse of AWS accounts for several organizations, Permiso found attackers had seized on stolen AWS credentials to interact with the large language models (LLMs) available on Bedrock. But they also soon discovered none of these AWS users had enabled logging (it is off by default), and thus they lacked any visibility into what attackers were doing with that access. So Permiso researchers decided to leak their own test AWS key on GitHub, while turning on logging so that they could see exactly what an attacker might ask for, and what the responses might be. Within minutes, their bait key was scooped up and used in a service that offers AI-powered sex chats online.

"After reviewing the prompts and responses it became clear that the attacker was hosting an AI roleplaying service that leverages common jailbreak techniques to get the models to accept and respond with content that would normally be blocked," Permiso researchers wrote in a report released today. "Almost all of the roleplaying was of a sexual nature, with some of the content straying into darker topics such as child sexual abuse," they continued. "Over the course of two days we saw over 75,000 successful model invocations, almost all of a sexual nature."

BleepingComputer's Ionut Ilascu reports: During a distributed denial-of-service campaign targeting organizations in the financial services, internet, and telecommunications sectors, volumetric attacks peaked at 3.8 terabits per second, the largest publicly recorded to date. The assault consisted of a "month-long" barrage of more than 100 hyper-volumetric DDoS attacks flooding the network infrastructure with garbage data. In a volumetric DDoS attack, the target is overwhelmed with large amounts of data to the point that they consume the bandwidth or exhaust the resources of applications and devices, leaving legitimate users with no access.

Many of the attacks aimed at the target's network infrastructure (network and transport layers L3/4) exceeded two billion packets per second (pps) and three terabits per second (Tbps). According to researchers at internet infrastructure company Cloudflare, the infected devices were spread across the globe but many of them were located in Russia, Vietnam, the U.S., Brazil, and Spain. The threat actor behind the campaign leveraged multiple types of compromised devices, which included a large number of Asus home routers, Mikrotik systems, DVRs, and web servers. Cloudflare mitigated all the DDoS attacks autonomously and noted that the one peaking at 3.8 Tbps lasted 65 seconds.