Daily Brief

Over 100 artificial intelligence (AI) and machine learning (ML) models on the Hugging Face platform were identified as malicious by JFrog security researchers. These models were found to contain code execution vulnerabilities through pickle files, potentially granting attackers full control of victims' machines through backdoors. The malicious models initiate reverse shell connections to an IP associated with the Korea Research Environment Open Network (KREONET) and potentially other IP addresses worldwide. Some authors discouraged downloading their own models, suggesting it could be a security demonstration, but the connection to an active IP crosses a line in security research ethics. The implications of this discovery reach beyond individual user risk to potential large-scale data breaches and corporate espionage. Researchers have also developed methods to prompt harmful responses from language models and a generative AI worm, Morris II, that can steal data and autonomously spread malware. The generative AI ecosystem's connectivity has been exploited to deliver malicious inputs to new applications in attacks comparable to buffer overflow and SQL injection techniques. This situation highlights the ongoing threat within open-source repositories and emphasizes the need for vigilance regarding the supply chain and generative AI services.

BleepingComputer uncovers over 60 domains impersonating major news outlets for content plagiarism and SEO manipulation. These fake news websites repost articles from reputable sources without attribution, deceiving readers and boosting their SEO. The operation, based in India, leverages this network to sell expensive advertorial slots to marketers under the guise of credible outlets. The network, likely promoting online gambling and betting, maintains a presence on Google News and social media platforms. Domains share common WordPress CMS, registrar, and host, with operations traced back to at least 2022. Connections are made to jackpotbetonline.com, a betting entity based in Gurugram, India, pointing to gambling promotion motives. The potential for evolving into a disinformation network remains, posing risks beyond trademark and copyright infringement.

U.S. agencies issue warnings on Phobos ransomware targeting critical U.S. sectors like government, emergency services, and healthcare. Phobos ransomware, structured as a Ransomware as a Service (RaaS), has several variants and employs tactics such as phishing and exploiting RDP services. The e-crime group behind Phobos uses Windows API functions, open-source tools, and sophisticated techniques to escalate privileges and maintain persistence. Attackers exfiltrate files before demanding ransom; after initial payment, 78% of organizations face renewed attacks, often with increased demands. Recently, Bitdefender reported a coordinated attack by CACTUS ransomware on two independent companies, exploiting a critical vulnerability within 24 hours of its disclosure. The median ransom demand in 2023 has risen to $600,000, with the average payment per victim reaching about $568,705 with no guarantee against recurring attacks.

Law enforcement recently disrupted LockBit ransomware operations, seizing the group’s website. LockBit reestablished a new site, listing ransom deadlines, including one involving data from Fulton County related to Donald Trump. Fulton County's ransom deadline passed without data release; LockBit alleges ransom payment, while officials deny any payment. Security analyst suggests data might have been seized by law enforcement rather than ransom being paid. Critical vulnerabilities reported in Cisco's NX-OS; patches recommended. CISA issued an advisory on Ivanti vulnerability mitigations which may not detect compromises; Ivanti recommends patching and using their Integrity Checker Tool. Researchers from Semperis warn of potential Silver SAML attacks allowing SAML token forgery without compromising ADFS servers, raising concerns similar to SolarWinds incident.

Ahead of Super Tuesday, FBI Director Christopher Wray and Senator Mark Warner highlighted the escalating threats to U.S. election security, particularly from foreign and domestic misinformation amplified by advances in AI technology. The barriers to creating realistic and convincing disinformation have lowered due to generative AI, potentially giving both sophisticated and less-sophisticated foreign adversaries new tools to influence elections. Homegrown criminals also pose a threat, as evidenced by incidents like an Alabama man threatening election workers and a political consultant hiring a magician to send fake robocalls. Officials express concern that the U.S. is less equipped to handle foreign intervention in the 2024 elections compared to 2020, with the spread of misinformation on social media as a central worry. AI-generated deepfakes and sophisticated bot farms are being used by nation-state actors to sow distrust in the electoral process and outcomes, further exacerbating the challenge of maintaining election integrity. The cost of creating deceptive content has become significantly affordable, raising concerns about the proliferation of such content and its potential to influence public perception and voter behavior.

A Linux backdoor named GTPDOOR has been discovered targeting mobile operator networks. GTPDOOR is believed to be associated with the threat group LightBasin (UNC1945), known for infiltrating telecommunications systems. The malware infiltrates systems critical to mobile communications like SGSN, GGSN, and P-GW, potentially accessing a telecom’s core network. Using the GPRS Tunnelling Protocol Control Plane, GTPDOOR camouflages its communications to avoid detection. It listens for specific "magic packets" to activate and perform operations, maintaining stealth through encrypted and authenticated packets. Antivirus engines struggle to detect GTPDOOR, as it targets outdated Linux versions and can masquerade its process name. Detection strategies include monitoring for abnormal socket activities and process names, with proposed defenses such as GTP firewalls and GSMA security guidelines.

BleepingComputer has uncovered a content farm operating over 60 websites that impersonate renowned media outlets such as the BBC, CNBC, and The Guardian. These fraudulent news websites, traced back to an operator in India, plagiarize content from legitimate sources without permission to boost their SEO and sell advertising. The fake news websites offer advertorial slots for press releases and product reviews, with prices ranging from $50 to $1000, potentially duping marketers seeking publicity. The syndicate of fake news sites may also be enrolled as a Google News publisher and maintains a social media presence to appear credible. BleepingComputer's ongoing investigation has revealed that while the current goal seems to be SEO optimization and ad sales; there's potential for the operation to spread disinformation in the future. The operation behind these websites is linked to promoting online gambling and betting activities and has been associated with jackpotbetonline.com, based in India. The content farm abuses trademarked media names which raises legal concerns, and the legitimacy of marketed products or services on these sites is questionable.

Hackers are using a new phishing kit named CryptoChameleon to target employees of the Federal Communications Commission (FCC) and users of cryptocurrency platforms (e.g., Binance, Coinbase, Kraken, Gemini). Attackers are employing a sophisticated social engineering strategy involving emails, SMS, and voice calls that mimic legitimate customer support, prompting victims to input sensitive information on fake SSO pages. The phishing campaign includes realistic replicas of Okta login screens and can adapt to request additional authentication, such as MFA codes, in real-time. Leveraging CAPTCHA challenges, the scam adds credibility to the phishing process and screens out non-human traffic. Victims are often redirected to the actual sign-in page or a decoy page claiming an account review to lower suspicion and provide attackers with more time to exploit the stolen data. Lookout's research revealed over 100 successful phishing victims, with many fake sites remaining active and continuing to harvest credentials hourly. Hosting for phishing pages transitioned from Hostwinds and Hostinger to Russia-based RetnNet, likely for prolonged operational capabilities for the scam sites. The identity of the threat actors remains unknown, as it's unclear whether CryptoChameleon is operated by a single group or shared among several cybercriminal entities.

Microsoft patched a serious Windows Kernel bug in February that was exploited as a zero-day since August 2023. The flaw, identified as CVE-2024-21338, affects Windows 10, Windows 11, and Windows Server versions since 2019. North Korean Lazarus hackers used the vulnerability to disable security tools and conduct stealth operations at the kernel level. The exploitation of the bug allows attackers to disrupt security software, hide malware indicators, and manipulate protected processes. Avast researchers linked the flaw to enhanced capabilities of the FudModule rootkit and a new remote access trojan used by Lazarus. Avast will share detailed insights into the attacks at BlackHat Asia in April. Users are urged to apply the latest security updates to protect against these sophisticated attacks by the Lazarus group.

A U.S. federal court has ruled that NSO Group must provide Meta with the source code for Pegasus spyware as part of ongoing litigation. The lawsuit, initiated by Meta in October 2019, accuses NSO Group of exploiting WhatsApp to install Pegasus on roughly 1,400 mobile devices, including those of Indian activists and journalists. NSO Group exploited a critical zero-day vulnerability in WhatsApp for the distribution of the spyware, which did not require call answer to infect the device. Though NSO Group must release details on the spyware, it is not required to disclose its server architecture or the identities of its clients. Amnesty International expressed disappointment that the clientele of NSO Group remains confidential, despite the firm facing U.S. sanctions for supplying cyber tools used in malicious operations against various individuals and entities. The order comes alongside revelations that the Intellexa Alliance's Predator mobile spyware is part of a new, complex delivery infrastructure involving multiple countries, highlighting ongoing concerns around mercenary spyware and its global implications.

The U.S. Department of Justice has unsealed an indictment against Iranian national Alireza Shafie Nasab for conducting a cyber campaign against U.S. entities. Nasab is accused of targeting over a dozen U.S. organizations, including government departments, defense contractors, and private firms since at least 2016. He purportedly used spear-phishing and custom applications to breach systems, deploy malware, and exfiltrate sensitive data. Techniques included impersonating individuals to gain the victim's confidence and leveraging compromised accounts to conduct further spear-phishing attacks. Nasab faces charges including wire fraud, conspiracy to commit computer and wire fraud, and aggravated identity theft, with a potential sentence of up to 47 years in prison. Despite Nasab's current fugitive status, the U.S. State Department is offering a reward of up to $10 million for information leading to his identification or location. The indictment links Nasab's activities to Mahak Rayan Afraz, a company with connections to Iran's Islamic Revolutionary Guard Corps and involved in previous social engineering campaigns.

Jack Teixeira, an Air National Guardsman, is expected to enter a guilty plea for leaking classified Pentagon documents. Teixeira shared top-secret files through Discord, which were then disseminated on social media. The leaked content suggests an obsession with mass shootings and conspiracy theories, and a possible attempt to impress online gamers. A US Air Force report implicated Teixeira as the sole individual responsible for the leak, but also pointed to oversight failures in his chain of command. Despite previously pleading not guilty, Teixeira has requested a court proceeding to change his plea; the specific charges he will plead guilty to are currently undisclosed. Teixeira's pre-trial arguments for release, drawing a comparison to former President Donald Trump's bail situation, were denied by a judge. The leaked documents contained sensitive information about geopolitical issues, including America's role in the Russia-Ukraine conflict and espionage developments in China. Following the incident, 15 Air National Guard leaders were disciplined, and the US Air Force has taken measures to strengthen classified data access protocols.

NSO Group, an Israeli company known for its Pegasus surveillance software, has been ordered by a U.S. federal judge to disclose the source code to Meta's WhatsApp, amidst allegations of unauthorized spying on 1,400 users. The court order, stemming from a 2019 lawsuit by WhatsApp, demands NSO Group to provide the source code of Pegasus and other relevant spyware created between April 29, 2018, and May 10, 2020. NSO Group has been accused of leveraging a vulnerability in WhatsApp's VoIP stack to allow remote access to victims' conversations and sensitive information. The legal setback allows NSO to keep its client list and server architecture details confidential. NSO Group declined to comment on the ruling. NSO Group, previously restructured in 2022, faces additional legal challenges from Apple and the Knight First Amendment Institute, with their immunity claims and attempts to dismiss lawsuits being rejected in U.S. courts. The U.S. has sanctioned NSO Group and similar spyware vendors, while the White House has issued an executive order limiting government use of such software, with certain exemptions. Amnesty International declares Pegasus spyware has been used against human rights defenders and journalists worldwide, implicating it in serious abuses, including the murder of Saudi journalist Jamal Khashoggi.

German authorities took down Crimemarket, the largest German-speaking cybercrime platform, arresting six individuals including one of its main operators. The platform, with over 180,000 users, facilitated the trade of illegal drugs, narcotics, and offered cybercrime services and criminal tutorials. The shutdown is the result of extensive investigations, with evidence gathered through a coordinated operation involving numerous searches. In North Rhine-Westphalia, where the main suspect was arrested, police seized 1 kilogram of marijuana, ecstasy tablets, and nearly 600,000 euros in cash and assets. Police have emphasized that the ongoing investigation targets not only the operators but also the users of the Crimemarket platform. User reports indicated prior accessibility issues on the site, which were later confirmed to be due to law enforcement actions rather than technical problems. The home page of Crimemarket remains online displaying a seizure notice, indicating long-term police monitoring and data confiscation as part of a Europe-wide operation.

Ransomware attacks have severely impacted healthcare institutions in recent months, with Change Healthcare, a UnitedHealth Group subsidiary, facing a significant attack linked to the BlackCat ransomware operation. The attack on Change Healthcare has disrupted billing services for pharmacies and patients, potentially affecting access to medications, with some patients being forced to pay high out-of-pocket costs. BlackCat claims to have stolen 6TB of data from Change Healthcare, putting millions of individuals' personal information at risk. A joint advisory from the FBI, CISA, and HHS warns of BlackCat's targeted attacks on U.S. hospitals, reflecting the seriousness of the threat. Rhysida ransomware operation is attempting to monetize stolen patient data from Lurie Children's Hospital in Chicago, demanding $3.6 million. Despite LockBit ransomware being hit by a police operation and returning with a reduced capacity and new infrastructure, it could potentially shut down after losing trust and tarnishing its reputation in the cybercrime community. Other ransomware activities include claims by an extortion group claiming to breach Epic Games without evidence of such an attack and multiple ransomware gangs exploiting ScreenConnect RCE vulnerability.