Search This Blog

Powered by Blogger.

Blog Archive

Labels

About Me

Showing posts with label Cloud. Show all posts

How ChatGPT prompt can allow cybercriminals to steal your Google Drive data


Chatbots and other AI tools have made life easier for threat actors. A recent incident highlighted how ChatGPT can be exploited to obtain API keys and other sensitive data from cloud platforms.

Prompt injection attacks leads to cloud access

Experts have discovered a new prompt injection attack that can turn ChatGPT into a hacker’s best friend in data thefts. Known as AgentFlayer, the exploit uses a single document to hide “secret” prompt instructions that target OpenAI’s chatbot. An attacker can share what appears to be a harmless document with victims through Google Drive, without any clicks.

Zero-click threat: AgentFlayer

AgentFlayer is a “zero-click” threat as it abuses a vulnerability in Connectors, for instance, a ChatGPT feature that connects the assistant to other applications, websites, and services. OpenAI suggests that Connectors supports a few of the world’s most widely used platforms. This includes cloud storage platforms such as Microsoft OneDrive and Google Drive.

Experts used Google Drive to expose the threats possible from chatbots and hidden prompts. 

GoogleDoc used for injecting prompt

The malicious document has a 300-word hidden malicious prompt. The text is size one, formatted in white to hide it from human readers but visible to the chatbot.

The prompt used to showcase AgentFlayer’s attacks prompts ChatGPT to find the victim’s Google Drive for API keys, link them to a tailored URL, and an external server. When the malicious document is shared, the attack is launched. The threat actor gets the hidden API keys when the target uses ChatGPT (the Connectors feature has to be enabled).

Othe cloud platforms at risk too

AgentFlayer is not a bug that only affects the Google Cloud. “As with any indirect prompt injection attack, we need a way into the LLM's context. And luckily for us, people upload untrusted documents into their ChatGPT all the time. This is usually done to summarize files or data, or leverage the LLM to ask specific questions about the document’s content instead of parsing through the entire thing by themselves,” said expert Tamir Ishay Sharbat from Zenity Labs.

“OpenAI is already aware of the vulnerability and has mitigations in place. But unfortunately, these mitigations aren’t enough. Even safe-looking URLs can be used for malicious purposes. If a URL is considered safe, you can be sure an attacker will find a creative way to take advantage of it,” Zenith Labs said in the report.

Vietnam Launches NDAChain for National Data Security and Digital Identity


Vietnam has launched NDAChain, a new blockchain network that allows only approved participants to join. The move is aimed at locking down Vietnam’s government data. 

About NDAChain

The network is built by the National Data Association and managed by the Ministry of Public Security’s Data Innovation and Exploitation Center. It will serve as the primary verification layer for tasks such as supply-chain logs, school transcripts, and hospital records.

According to experts, NDAChain is based on a hybrid model, relying on a Proof-of-Authority mechanism to ensure only authorized nodes can verify transactions. It also adds Zero-Knowledge-Proofs to protect sensitive data while verifying its authenticity. According to officials, NDAChain can process between 1,200 and 3,600 transactions per second, a statistic that aims to support faster verifications in logistics, e-government, and other areas. 

Two new features

The networks have two main features: NDA DID offers digital IDs that integrate with Vietnam’s current VNeID framework, allowing users to verify their IDs online when signing documents or using services. On the other hand, NDATrace provides end-to-end product tracking via GS1 and EBSI Trace standards. Items are tagged with unique identifiers that RFID chips or QR codes can scan, helping businesses prove verification to overseas procurers and ease recalls in case of problems.

Privacy layer and network protection

NDAChain works as a “protective layer” for Vietnam’s digital infrastructure, built to scale as data volume expands. Digital records can be verified without needing personal details due to the added privacy tools. The permissioned setup also offers authorities more control over people joining the network. According to reports, total integration with the National Data Center will be completed by this year. The focus will then move towards local agencies and universities, where industry-specific Layer 3 apps are planned for 2026.

According to Vietnam Briefing, "in sectors such as food, pharmaceuticals, and health supplements, where counterfeit goods remain a persistent threat, NDAChain enables end-to-end product origin authentication. By tracing a product’s whole journey from manufacturer to end-consumer, businesses can enhance brand trust, reduce legal risk, and meet rising regulatory demands for transparency."

Why Running AI Locally with an NPU Offers Better Privacy, Speed, and Reliability

 

Running AI applications locally offers a compelling alternative to relying on cloud-based chatbots like ChatGPT, Gemini, or Deepseek, especially for those concerned about data privacy, internet dependency, and speed. Though cloud services promise protections through subscription terms, the reality remains uncertain. In contrast, using AI locally means your data never leaves your device, which is particularly advantageous for professionals handling sensitive customer information or individuals wary of sharing personal data with third parties. 

Local AI eliminates the need for a constant, high-speed internet connection. This reliable offline capability means that even in areas with spotty coverage or during network outages, tools for voice control, image recognition, and text generation remain functional. Lower latency also translates to near-instantaneous responses, unlike cloud AI that may lag due to network round-trip times. 

A powerful hardware component is essential here: the Neural Processing Unit (NPU). Typical CPUs and GPUs can struggle with AI workloads like large language models and image processing, leading to slowdowns, heat, noise, and shortened battery life. NPUs are specifically designed for handling matrix-heavy computations—vital for AI—and they allow these models to run efficiently right on your laptop, without burdening the main processor. 

Currently, consumer devices such as Intel Core Ultra, Qualcomm Snapdragon X Elite, and Apple’s M-series chips (M1–M4) come equipped with NPUs built for this purpose. With one of these devices, you can run open-source AI models like DeepSeek‑R1, Qwen 3, or LLaMA 3.3 using tools such as Ollama, which supports Windows, macOS, and Linux. By pairing Ollama with a user-friendly interface like OpenWeb UI, you can replicate the experience of cloud chatbots entirely offline.  

Other local tools like GPT4All and Jan.ai also provide convenient interfaces for running AI models locally. However, be aware that model files can be quite large (often 20 GB or more), and without NPU support, performance may be sluggish and battery life will suffer.  

Using AI locally comes with several key advantages. You gain full control over your data, knowing it’s never sent to external servers. Offline compatibility ensures uninterrupted use, even in remote or unstable network environments. In terms of responsiveness, local AI often outperforms cloud models due to the absence of network latency. Many tools are open source, making experimentation and customization financially accessible. Lastly, NPUs offer energy-efficient performance, enabling richer AI experiences on everyday devices. 

In summary, if you’re looking for a faster, more private, and reliable AI workflow that doesn’t depend on the internet, equipping your laptop with an NPU and installing tools like Ollama, OpenWeb UI, GPT4All, or Jan.ai is a smart move. Not only will your interactions be quick and seamless, but they’ll also remain securely under your control.

Security Breach Reveals "Catwatchful" Spyware is Snooping on Users

Security Breach Reveals "Catwatchful" Spyware is Snooping on Users

A security bug in a stealthy Android spyware operation, “Catwatchful,” has exposed full user databases affecting its 62,000 customers and also its app admin. The vulnerability was found by cybersecurity expert Eric Daigle reported about the spyware app’s full database of email IDs and plaintext passwords used by Catwatchful customers to access stolen data from the devices of their victims. 

Most of the victims were based in India, Argentina, Peru, Mexico, Colombia, Bolivia, and Ecuador. A few records date back to 2018. The leaked database also revealed the identity of the Catwatchful admin called Omar Soca Char.

The Catwatchful database also revealed the identity of the spyware operation’s administrator, Omar Soca Charcov, a developer based in Uruguay.

About Catwatchful

Catwatchful is a spyware that pretends to be a child monitoring app, claiming to be “invisible and can not be detected,” while it uploads the victim’s data to a dashboard accessible to the person who planted the app. The stolen data includes real-time location data, victims’ photos, and messages.  The app can also track live ambient audio from the device’s mic and access the phone camera (both front and rear).

Catwatchful and similar apps are banned on app stores, and depend on being downloaded and deployed by someone having physical access to a victim’s phone. These apps are famous as “stalkerware” or “spouseware” as they are capable of unauthorized and illegal non-consensual surveillance of romantic partners and spouses. 

Rise of spyware apps

The Catwatchful incident is the fifth and latest in this year’s growing list of stalkerware scams that have been breached, hacked, or had their data exposed. 

How was the spyware found?

Daigle has previously discovered stalkerware exploits. Catwatchful uses a custom-made API, which the planted app uses to communicate to send data back to Catwatchful servers. The stalkerware also uses Google Firebase to host and store stolen data. 

According to Techradar, the “data was stored on Google Firebase, sent via a custom API that was unauthenticated, resulting in open access to user and victim data. The report also confirms that, although hosting had initially been suspended by HostGator, it had been restored via another temporary domain."

Amid Federal Crackdown, Microsoft Warns Against Rising North Korean Jobs Scams

Amid Federal Crackdown, Microsoft Warns Against Rising North Korean Jobs Scams

North Korean hackers are infiltrating high-profile US-based tech firms through scams. Recently, they have even advanced their tactics, according to the experts. In a recent investigation by Microsoft, the company has requested its peers to enforce stronger pre-employment verification measures and make policies to stop unauthorized IT management tools. 

Further investigation by the US government revealed that these actors were working to steal money for the North Korean government and use the funds to run its government operations and its weapons program.  

US imposes sanctions against North Korea

The US has imposed strict sanctions on North Korea, which restrict US companies from hiring North Korean nationals. It has led to threat actors making fake identities and using all kinds of tricks (such as VPNs) to obscure their real identities and locations. This is being done to avoid getting caught and get easily hired. 

Recently, the threat actors have started using spoof tactics such as voice-changing tools and AI-generated documents to appear credible. In one incident, the scammers somehow used an individual residing in New Jersey, who set up shell companies to fool victims into believing they were paying a legitimate local business. The same individual also helped overseas partners to get recruited. 

DoJ arrests accused

The clever campaign has now come to an end, as the US Department of Justice (DoJ) arrested and charged a US national called Zhenxing “Danny” Wanf with operating a “year-long” scam. The scheme earned over $5 million. The agency also arrested eight more people - six Chinese and two Taiwanese nationals. The arrested individuals are charged with money laundering, identity theft, hacking, sanctions violations, and conspiring to commit wire fraud.

In addition to getting paid in these jobs, which Microsoft says is a hefty payment, these individuals also get access to private organization data. They exploit this access by stealing sensitive information and blackmailing the company.

Lazarus group behind such scams

One of the largest and most infamous hacking gangs worldwide is the North Korean state-sponsored group, Lazarus. According to experts, the gang extorted billions of dollars from the Korean government through similar scams. The entire campaign is popular as “Operation DreamJob”. 

"To disrupt this activity and protect our customers, we’ve suspended 3,000 known Microsoft consumer accounts (Outlook/Hotmail) created by North Korean IT workers," said Microsoft.

Telegram’s Alleged Ties to Russian Intelligence Raise Global Surveillance Fears

 

A new investigation by Russian media outlet Important Stories, in collaboration with the Organized Crime and Corruption Reporting Project (OCCRP), has sparked fresh scrutiny over Telegram’s connections to Russia’s intelligence services. The popular messaging platform, long regarded for its privacy features, may have indirect links to the Russian Federal Security Service (FSB), raising significant concerns for users worldwide.

At the center of the probe is a company called Global Network Management (GNM), which plays a critical role in routing Telegram’s messages. Although GNM is officially incorporated in the Caribbean nation of Antigua and Barbuda, it operates primarily from Russia. Its owner, Vladimir Vedeneev, is a Russian engineer with long-standing ties to Telegram founder Pavel Durov. Legal filings show that Vedeneev is the only individual authorized to manage certain Telegram servers, including those based in the U.S. 

Vedeneev also runs other firms—such as Globalnet and Electrontelecom—that reportedly supply telecommunications infrastructure to various Russian state entities, including the FSB. These companies have been linked to classified government projects involving surveillance and defense. 

The IP addresses used by Telegram used to be owned by Russian firms with FSB affiliations. These IPs still appear to be registered in Russia, and might be responsible for allowing user activity to be traced back through Russian-controlled networks. Telegram users typically rely on regular cloud chats, which—unlike its secret chats—are not end-to-end encrypted and are stored on Telegram’s servers. Security analysts warn that if Vedeneev’s companies manage routing systems and network infrastructure, they could potentially access user metadata, including IP addresses, device IDs, and location data. 

Though message content may remain encrypted, this metadata could still be exploited for surveillance. Moreover, Telegram transmits unique device identifiers in an unencrypted format, creating additional vulnerability. Experts caution that Russian intelligence could leverage this data to monitor users, particularly dissidents, journalists, or foreign nationals viewed as threats. Telegram has refuted the claims, stating that it has no employees or servers in Russia and that its infrastructure remains fully under the control of its internal teams. 

The company maintains that no third party, including vendors, can access confidential user data or systems. However, Telegram has yet to directly address the investigation’s core claims regarding GNM, Vedeneev, or the related infrastructure providers. The platform also hasn’t explained how it protects users if server operators have potential intelligence ties or why certain data is still sent without encryption. 

The issue is especially relevant in Ukraine, where Telegram has over 10 million users and is a major source of news and official communication. While President Volodymyr Zelensky’s administration uses the app for public updates, growing concerns around disinformation and espionage have prompted discussions about its continued use. 

As the investigation raises critical questions about the app’s security, the broader implications for global digital privacy and national security remain in sharp focus.

WhatsApp Reveals "Private Processing" Feature for Cloud Based AI Features

WhatsApp Reveals "Private Processing" Feature for Cloud Based AI Features

WhatsApp claims even it can not process private data

WhatsApp has introduced ‘Private Processing,’ a new tech that allows users to use advanced AI features by offloading tasks to privacy-preserving cloud servers, without exposing their chat to Meta. Meta claims even it cannot see the messages while processing them. The system employs encrypted cloud infrastructure and hardware-based isolation without making it visible to anyone, even Meta, or processing data. 

About private processing

For those who decide to use Private Processing, the system works in an obscure verification via the user’s WhatsApp client to confirm the user’s validity. 

Meta claims this system keeps WhatsApp’s end-to-end encryption intact while offering AI features in chats. However, the feature currently applies only to select use cases and excludes Meta’s broader AI deployments, including those used in India’s public service systems.

Private processing employs Trusted Execution Environments (TEEs) — safe virtual machines that use cloud infrastructure to keep AI requests hidden. 

About the system

  • Performs encryption of user requests from the system to the TEE utilizing end-to-end encryption
  • Encrypts user requests from the device to the TEE using end-to-end encryption
  • Restricts storage or logging of messages post-processing
  • Reports logs and binary images for external verification and audits

WhatsApp builds AI through wider privacy concerns

According to Meta, the Private processing is a response to privacy questions around AI and messaging. WhatsApp has now joined other companies like Apple that have introduced confidential AI computing models in the previous year. “To validate our implementation of these and other security principles, independent security researchers will be able to continuously verify our privacy and security architecture and its integrity,” Meta said.

It is similar to Apple’s private cloud computing in terms of public transparency and stateless processing. Currently, however, WhatsApp is using them only for select features. Apple, on the other hand, has declared plans to implement this model throughout all its AI tools, whereas WhatsApp has not made such claims, yet. 

WhatsApp says, “Private Processing uses anonymous credentials to authenticate users over OHTTP. This way, Private Processing can authenticate users to the Private Processing system but remains unable to identify them.”

Posture Management Emerges as Strategic Cybersecurity Priority Amid Cloud and Data Fragmentation

 

Posture management is rapidly evolving into a cornerstone of enterprise cybersecurity as organizations grapple with increasing digital complexity. With infrastructures now sprawling across cloud platforms, identity services, and data environments, the traditional model of siloed risk monitoring is no longer sustainable. As a result, cybersecurity leaders are embracing posture management not only to gauge exposure but also to orchestrate defenses in real time. 

This shift reflects a broader industry movement toward unifying visibility and control. “From a business perspective, large organizations have M&A — they have rollups; they have multiple divisions. They’re not centralized; they’re across globes,” said Erik Bradley, chief strategist and director of research at Enterprise Technology Research. “There’s no way that we’re ever going to see a consolidation on one platform.” 

Bradley shared these insights during a conversation with theCUBE’s Jon Oltsik and Dave Vellante at the RSAC 2025 Conference, hosted by SiliconANGLE Media. The discussion focused on how posture management is becoming integral to modern security operations by improving visibility, minimizing tool sprawl, and enabling strategic risk reduction across complex IT environments. Security teams are increasingly recognizing the limitations of point solutions. 

Instead, they’re exploring how posture management can serve as a foundational layer across enterprise-wide platforms. “We’re carving up terminology and confusing the market,” said Oltsik. “IT is moving so quickly and it’s so specialized that you need specialized posture management tools for cloud, identity, and data.” Leading cybersecurity vendors like CrowdStrike and Palo Alto Networks are embedding posture management into broader security suites, aligning it with automation, identity access controls, and even password management. 

These integrations aim to reduce operational overhead while enabling faster, more accurate threat detection. According to Bradley, these vendors view posture management as both a preventive control in peacetime and a readiness tool in active cyber warfare scenarios. However, challenges persist—particularly around data fragmentation. Although many vendors tout strong telemetry capabilities, few offer complete visibility across all domains. 

This leaves enterprises vulnerable to gaps in their defenses, especially as they try to consolidate vendors and reduce redundancy. “No CSO is going to go all-in with one provider,” Bradley emphasized. “They’re focused on consolidating redundant vendors and streamlining operations without sacrificing visibility or security.” 

As cybersecurity evolves, posture management is no longer a niche function—it’s becoming the backbone of a resilient, scalable defense strategy.

Microsoft Alerts Users About Password-spraying Attack

Microsoft Alerts Users About Password-spraying Attack

Microsoft alerts users about password-spraying attacks

Microsoft has warned users about a new password-spraying attack by a hacking group Storm-1977 that targets cloud users. The Microsoft Threat Intelligence team reported a new warning after discovering threat actors are abusing unsecured workload identities to access restricted resources. 

According to Microsoft, “Container technology has become essential for modern application development and deployment. It's a critical component for over 90% of cloud-native organizations, facilitating swift, reliable, and flexible processes that drive digital transformation.” 

Hackers use adoption-as-a-service

Research says 51% of such workload identities have been inactive for one year, which is why attackers are exploiting this attack surface. The report highlights the “adoption of containers-as-a-service among organizations rises.” According to Microsoft, it continues to look out for unique security dangers that affect “containerized environments.” 

The password-spraying attack targeted a command line interface tool “AzureChecker” to download AES-encrypted data which revealed the list of password-spray targets after it was decoded. To make things worse, the “threat actor then used the information from both files and posted the credentials to the target tenants for validation.”

The attack allowed the Storm-1977 hackers to leverage a guest account to make a compromised subscription resource group and over 200 containers that were used for crypto mining. 

Mitigating password-spraying attacks

The solution to the problem of password spraying attacks is eliminating passwords. It can be done by moving towards passkeys, a lot of people are already doing that. 

Microsoft has suggested these steps to mitigate the issue

  • Use strong authentication while putting sensitive interfaces to the internet. 
  • Use strong verification methods for the Kubernetes API to stop hackers from getting access to the cluster even when valid credentials like kubeconfig are obtained.  
  • Don’t use the read-only endpoint of Kubelet on port 10255, which doesn’t need verification. 

Modify the Kubernetes role-based access controls for every user and service account to only retain permissions that are required. 

According to Microsoft, “Recent updates to Microsoft Defender for Cloud enhance its container security capabilities from development to runtime. Defender for Cloud now offers enhanced discovery, providing agentless visibility into Kubernetes environments, tracking containers, pods, and applications.” These updates upgrade security via continuous granular scanning. 

Hitachi Vantara Takes Servers Offline Following Akira Ransomware Attack

 

Hitachi Vantara, a subsidiary of Japan's Hitachi conglomerate, temporarily shut down several servers over the weekend after falling victim to a ransomware incident attributed to the Akira group.

The company, known for offering data infrastructure, cloud operations, and cyber resilience solutions, serves government agencies and major global enterprises like BMW, Telefónica, T-Mobile, and China Telecom.

In a statement to BleepingComputer, Hitachi Vantara confirmed the cyberattack and revealed it had brought in external cybersecurity specialists to assess the situation. The company is now working to restore all affected systems.

“On April 26, 2025, Hitachi Vantara experienced a ransomware incident that has resulted in a disruption to some of our systems," Hitachi Vantara told BleepingComputer.

"Upon detecting suspicious activity, we immediately launched our incident response protocols and engaged third-party subject matter experts to support our investigation and remediation process. Additionally, we proactively took our servers offline in order to contain the incident.

We are working as quickly as possible with our third-party subject matter experts to remediate this incident, continue to support our customers, and bring our systems back online in a secure manner. We thank our customers and partners for their patience and flexibility during this time."

Although the company has not officially attributed the breach to any specific threat actor, BleepingComputer reports that sources have linked the attack to the Akira ransomware operation. Insiders allege that the attackers exfiltrated sensitive data and left ransom notes on infiltrated systems.

While cloud services remained unaffected, sources noted that internal platforms at Hitachi Vantara and its manufacturing arm experienced disruption. Despite these outages, clients operating self-hosted systems are still able to access their data.

A separate source confirmed that several government-led initiatives have also been impacted by the cyberattack.

Akira ransomware first appeared in March 2023 and swiftly became notorious for targeting a wide range of sectors worldwide. Since its emergence, the group has reportedly compromised more than 300 organizations, including high-profile names like Stanford University and Nissan (in Oceania and Australia).

The FBI estimates that Akira collected over $42 million in ransom payments by April 2024 after infiltrating over 250 organizations. According to chat logs reviewed by BleepingComputer, the gang typically demands between $200,000 and several million dollars, depending on the scale and sensitivity of the targeted entity.

Keywords: ransomware, cybersecurity, Hitachi, Akira, cloud, breach, data, FBI, malware, attack, encryption, extortion, hacking, disruption, recovery, infrastructure, digital, protection

New Report Reveals Hackers Now Aim for Money, Not Chaos

New Report Reveals Hackers Now Aim for Money, Not Chaos

Recent research from Mandiant revealed that financially motivated hackers are the new trend, with more than (55%) of criminal gangs active in 2024 aiming to steal or extort money from their targets, a sharp rise compared to previous years. 

About the report

The main highlight of the M-Trends report is that hackers are using every opportunity to advance their goals, such as using infostealer malware to steal credentials. Another trend is attacking unsecured data repositories due to poor security hygiene. 

Hackers are also exploiting fractures and risks that surface when an organization takes its data to the cloud. “In 2024, Mandiant initiated 83 campaigns and five global events and continued to track activity identified in previous years. These campaigns affected every industry vertical and 73 countries across six continents,” the report said. 

Ransomware-related attacks accounted for 21% of all invasions in 2024 and comprised almost two-thirds of cases related to monetization tactics. This comes in addition to data theft, email hacks, cryptocurrency scams, and North Korean fake job campaigns, all attempting to get money from targets. 

Exploits were amid the most popular primary infection vector at 33%, stolen credentials at 16%, phishing at 14%, web compromises at 9%, and earlier compromises at 8%. 

Finance in danger

Finance topped in the targeted industry, with more than 17% of attacks targeting the sector, followed closely by professional services and business (11%), critical industries such as high tech (10%), governments (10%), and healthcare (9%). 

Experts have highlighted a broader target of various industries, suggesting that anyone can be targeted by state-sponsored attacks, either politically or financially motivated.  

Stuart McKenzie, Managing Director, Mandiant Consulting EMEA. said “Financially motivated attacks are still the leading category. “While ransomware, data theft, and multifaceted extortion are and will continue to be significant global cybercrime concerns, we are also tracking the rise in the adoption of infostealer malware and the developing exploitation of Web3 technologies, including cryptocurrencies.” 

He also stressed that the “increasing sophistication and automation offered by artificial intelligence are further exacerbating these threats by enabling more targeted, evasive, and widespread attacks. Organizations need to proactively gather insights to stay ahead of these trends and implement processes and tools to continuously collect and analyze threat intelligence from diverse sources.”

Malware Campaign Uses Fake CAPTCHAs, Tricks Online Users


Researchers at Netskope Threat Labs have found a new malicious campaign that uses tricky tactics to distribute the Legion Loader malware. The campaign uses fake CAPTCHAs and CloudFlare Turnstile to trap targets into downloading malware that leads to the installation of malicious browser extensions. 

Malware campaign attacks users via fake CAPTCHAs

The hackers have attacked over 140 Netskope customers situated in Asia, North America, and Southern Europe throughout different segments, driven by the financial and tech sectors. 

Netskope has been examining different phishing and malware campaigns targeting users who look for PDF documents online. Hackers use tricky ways within these PDFs to resend victims to malicious websites or lure them into downloading malware. In the newly found campaign, they used fake CAPTCHAs and Cloudflare Turnstile to distribute the LegionLoader payload. 

Important stages in the attack chain

The infection begins with a drive-by download when a target looks for a particular document and is baited to a malicious site.

The downloaded file contains a fake CAPTCHA. If clicked, it redirects the user via a Clloudfare Turnstile CAPTCHA to a notification page. 

In the last step, victims are urged to allow browser notifications.

Attack tactic in detail

When a user blocks the browser notification prompt or uses a browser that doesn’t support notifications, they are redirected to download harmless apps like Opera or 7-Zip. However, if the user agrees to receive browser notifications, they are redirected to another Cloudflare Turnstile CAPTCHA. Once this is done, they are sent to a page with instructions on how to download their file.

The download process requires the victim to open the Windows Run window (win + r) and put content copied to the clipboard (ctrl + v), and “ execute it by pressing enter (we described a similar approach in a post about Lumma Stealer),” Netscope said. In this incident, the command in the clipboard uses the “ command prompt to run cURL and download an MSI file.” After this, the “command opens File Explorer, where the MSI file has been downloaded. When the victim runs the MSI file, it will execute the initial payload.”

Hackers use different tactics to avoid getting caught

To avoid detection, the campaign uses a legitimate VMware-signed app that sideloads a malicious DLL to run and load the LegionLeader payload. Later, a new custom algorithm is used to remove the LegionLeader shellcode loader. 

In the final stage, the hackers install a malicious browser extension that can steal sensitive info across different browsers, such as Opera, Chrome, Brave, and Edge. Netscope warns of an alarming trend where hackers are targeting users searching for PDF docs online via sophisticated tactics to install malware.

Alibaba Launches Latest Open-source AI Model from Qwen Series for ‘Cost-effective AI agents’

Alibaba Launches Lates Open-source AI Model from Qwen Series for ‘Cost-effective AI agents’

Last week, Alibaba Cloud launched its latest AI model in its “Qwen series,” as large language model (LLM) competition in China continues to intensify after the launch of famous “DeepSeek” AI.

The latest "Qwen2.5-Omni-7B" is a multimodal model- it can process inputs like audio/video, text, and images- while also creating real-time text and natural speech responses, Alibaba’s cloud website reports. It also said that the model can be used on edge devices such as smartphones, providing higher efficiency without giving up on performance. 

According to Alibaba, the “unique combination makes it the perfect foundation for developing agile, cost-effective AI agents that deliver tangible value, especially intelligent voice applications.” For instance, the AI can be used to assist visually impaired individuals to navigate their environment via real-time audio description. 

The latest model is open-sourced on forums GitHub and Hugging Face, after a rising trend in China post DeepSeek breakthrough R1 model open-source. Open-source means a software in which the source code is created freely on web for potential modification and redistribution. 

In recent years, Alibaba claims it has open-sourced more that 200 generative AI models. In the noise of China’s AI dominance intensified by DeepSeek due to its shoe string budget and capabilities, Alibaba and genAI competitors are also releasing new, cost-cutting models and services an exceptional case.

Last week, Chinese tech mammoth Baidu launched a new multimodal foundational model and its first reasoning-based model. Likewise, Alibaba introduced its updated Qwen 2.5 AI model in January and also launched a new variant of its AI assistant tool Quark this month. 

Alibaba has also made strong commitments to its AI plan, recently, it announced a plan to put $53 billion in its cloud computing and AI infrastructure over the next three years, even surpassing its spending in the space over the past decade. 

CNBC talked with Kai Wang, Asia Senior equity analyst at Morningstar, Mr Kai told CNBC that “large Chinese tech players such as Alibaba, which build data centers to meet the computing needs of AI in addition to building their own LLMs, are well positioned to benefit from China's post-DeepSeek AI boom.” According to CNBC, “Alibaba secured a major win for its AI business last month when it confirmed that the company was partnering with Apple to roll out AI integration for iPhones sold in China.”

Google Cloud Introduces Quantum-Safe Digital Signatures

 

As quantum computing advances, Google Cloud is taking a significant step toward securing its platform against future threats. The company has announced the introduction of quantum-safe digital signatures in its Cloud Key Management Service (KMS), currently available in preview. 

This move is part of a broader initiative to prepare for the potential risks that quantum computers pose to modern encryption systems. While fully capable quantum computers are not expected to be widely available for at least a decade, they could one day break most of today’s encryption methods in a matter of hours. This looming possibility has led to concerns over a harvest-now-decrypt-later strategy employed by cybercriminals. 

In this method, attackers steal encrypted data today, intending to decrypt it once quantum computing becomes powerful enough. To counter this risk, researchers are developing post-quantum cryptography (PQC)—encryption techniques specifically designed to withstand quantum attacks. One major security risk posed by quantum computing is the potential forgery and manipulation of digital signatures. 

Digital signatures authenticate documents and communications, ensuring they have not been tampered with. If compromised, they could allow attackers to impersonate legitimate users, forge transactions, or spread malware under trusted identities. Google Cloud recognizes the importance of addressing these concerns early and has introduced quantum-resistant digital signatures to build a more secure infrastructure. 

This initiative also aims to set an industry precedent for other cloud service providers. As part of its commitment to transparency and security, Google Cloud has announced that its quantum-related cryptographic implementations will be included in its open-source cryptographic libraries, BoringCrypto and Tink. This allows security researchers and developers to review, audit, and contribute to these implementations, ensuring their robustness against potential threats. 

The new quantum-safe digital signatures in Cloud KMS specifically implement ML-DSA-65 and SLH-DSA-SHA2-128S, two PQC algorithms that adhere to NIST (National Institute of Standards and Technology) standards. Google Cloud has also confirmed plans to integrate additional PQC algorithms into its Hardware Security Modules (HSMs), which are specialized devices designed to provide extra layers of cryptographic security.  

By rolling out these quantum-resistant digital signatures, Google Cloud is giving customers the opportunity to test PQC algorithms in Cloud KMS and provide feedback on their performance and integration. This allows businesses to prepare for a post-quantum future, ensuring their data remains secure even as computing power evolves. 

Google Cloud sees this initiative as a crucial first step toward a fully quantum-resistant cloud ecosystem, demonstrating its dedication to staying ahead of emerging cybersecurity challenges.

User Tracking: Google to Store User Data for 180 Days

User Tracking: Google Announces to Store User Data for 180 Days

Google has made a major change in its user tracking, a big leap in privacy concerns for users. Google will stop the nosy cloud storage of data it gets from tracking user location in real time. 

The privacy change

Called Google Maps Timeline, from December, Google will save user location data for a maximum of 180 days. After the duration ends, the data will be erased from Google Cloud servers. 

The new policy means Google can only save a user’s movements and whereabouts for 6 months, the user has an option to store the data on a personal device, but the cloud data will be permanently deleted from Google servers.

The new privacy change is welcomed, smartphones can balance privacy and convenience in terms of data storage, but nothing is more important than location data

Users can change settings that suit them best, but the majority go with default settings. The problem here arises when Google uses user data for suggesting insights (based on anonymous location data), or improving Google services like ads products.

Why important 

The Google Maps Timeline feature addresses questions about data privacy and security. The good things include:

Better privacy: By restricting the storage timeline of location data on the cloud, Google can reduce data misuse. Limiting the storage duration means less historical data is exposed to threat actors if there's a breach.

More control to users: When users have the option to retain location data on their devices, it gives them ownership over their personal data. Users can choose whether to delete their location history or keep the data.

Accountability from Google: The move is a positive sign toward building transparency and trust, showing a commitment to user privacy. 

How will it impact users?

Services: Google features that use location history data for tailored suggestions might be impacted, and users may observe changes in correct location-based suggestions and targeted ads. 

The problem in data recovery: For users who like to store their data for a longer duration, the new move can be a problem. Users will have to self-back up data if they want to keep it for more than 180 days. 

The Rise of Manual Techniques in Ransomware Attacks: A Growing Threat

The Rise of Manual Techniques in Ransomware Attacks: A Growing Threat

A recent report by CrowdStrike observes on a disturbing trend: the increasing use of manual techniques in ransomware attacks. This shift towards hands-on-keyboard activities is not only making these attacks more sophisticated but also more challenging to detect and mitigate.

The Surge in Interactive Intrusions

According to CrowdStrike’s findings, there has been a staggering 55% increase in interactive intrusions over the past year. These intrusions, characterized by direct human involvement rather than automated scripts, account for nearly 90% of e-crime activities. This trend underscores a critical shift in the tactics employed by cybercriminals, who are now leveraging manual techniques to bypass traditional security measures and achieve their malicious objectives.

Why Manual Techniques?

The adoption of manual techniques in ransomware attacks offers several advantages to cybercriminals. Firstly, these techniques allow attackers to adapt and respond in real-time to the defenses they encounter. Unlike automated attacks, which follow predefined scripts, manual intrusions enable attackers to think on their feet, making it harder for security systems to predict and counter their moves.

Secondly, manual techniques often involve the use of legitimate tools and credentials, making it difficult for security teams to distinguish between malicious and benign activities. This tactic, known as “living off the land,” involves using tools that are already present in the target environment, such as PowerShell or Remote Desktop Protocol (RDP). By blending in with normal network traffic, attackers can evade detection for extended periods, increasing the likelihood of a successful attack.

The Impact on the Technology Sector

The technology sector has been particularly hard-hit by this surge in manual ransomware attacks. CrowdStrike’s report indicates a 60% rise in such attacks targeting tech companies. This sector is an attractive target for cybercriminals due to its vast repositories of sensitive data and intellectual property. Additionally, technology companies often have complex and interconnected systems, providing multiple entry points for attackers to exploit.

The consequences of a successful ransomware attack on a tech company can be devastating. Beyond the immediate financial losses from ransom payments, these attacks can lead to prolonged downtime, loss of customer trust, and significant reputational damage. In some cases, the recovery process can take months, further compounding the financial and operational impact.

What to do?

Enhanced Monitoring and Detection: Implement advanced monitoring tools that can detect anomalous behavior indicative of manual intrusions. Behavioural analytics and machine learning can help identify patterns that deviate from the norm, providing early warning signs of an attack.

Regular Security Training: Educate employees about the latest phishing techniques and social engineering tactics used by cybercriminals. Regular training sessions can help staff recognize and report suspicious activities, reducing the risk of initial compromise.

Zero Trust Architecture: Adopt a Zero Trust approach to security, where no user or device is trusted by default. Implement strict access controls and continuously verify the identity and integrity of users and devices accessing the network.

Incident Response Planning: Develop and regularly update an incident response plan that outlines the steps to take in the event of a ransomware attack. Conduct regular drills to ensure that all team members are familiar with their roles and responsibilities during an incident.

Backup and Recovery: Maintain regular backups of critical data and ensure that these backups are stored securely and inaccessible from the main network. Regularly test the recovery process to ensure that data can be restored quickly in the event of an attack.

Moroccan Cybercrime Group Storm-0539 Exploits Gift Card Systems with Advanced Phishing Attacks

 

A Morocco-based cybercrime group, Storm-0539, is making headlines for its sophisticated email and SMS phishing attacks aimed at stealing and reselling gift cards. Microsoft's latest Cyber Signals report reveals that this group is responsible for significant financial theft, with some companies losing up to $100,000 daily. 

First identified by Microsoft in December 2023, Storm-0539, also known as Atlas Lion, has been active since late 2021. The group employs social engineering techniques to harvest victims' credentials through adversary-in-the-middle (AitM) phishing pages. They exploit this access to register their own devices, bypass authentication, and maintain persistent access to create fraudulent gift cards. 

The group's attack strategy includes gaining covert access to cloud environments for extensive reconnaissance, targeting large retailers, luxury brands, and fast-food chains. They aim to redeem and sell gift cards on black markets or use money mules to cash out. This marks an evolution from their previous tactics of stealing payment card data via malware on point-of-sale (PoS) devices. 

Microsoft noted a 30% increase in Storm-0539's activities between March and May 2024, emphasizing their deep understanding of cloud systems to manipulate gift card issuance processes. In addition to stealing login credentials, Storm-0539 targets secure shell (SSH) passwords and keys, which are either sold or used for further attacks. The group uses internal company mailing lists to send phishing emails, enhancing their credibility and sets up new phishing websites by exploiting free trial or student accounts on cloud platforms. 

The FBI has warned about Storm-0539's smishing attacks on retail gift card departments, using sophisticated phishing kits to bypass multi-factor authentication (MFA). The group's ability to adapt and pivot tactics after detection underscores their persistence and resourcefulness. Microsoft urges companies to monitor gift card portals closely and implement conditional access policies to strengthen security. They highlight the effectiveness of using additional identity-driven signals, such as IP address and device status, alongside MFA. 

Meanwhile, Enea researchers have identified broader criminal campaigns exploiting cloud storage services like Amazon S3 and Google Cloud Storage for SMS-based gift card scams. These scams use legitimate-looking URLs to bypass firewalls and redirect users to malicious websites that steal sensitive information. 

Storm-0539's operations exemplify the increasing sophistication of financially motivated cybercriminals, borrowing techniques from state-sponsored actors to remain undetected. As these threats evolve, robust cybersecurity measures and vigilant monitoring are crucial to protect sensitive information and financial assets.

Empowering Indigenous Data Sovereignty: The TTP-Microsoft Partnership

 

The recent partnership between Te Tumu Paeroa (TTP), the office of the Māori Trustee, and Microsoft for the forthcoming data centres in Aotearoa New Zealand marks a groundbreaking development with potential global implications for indigenous data sovereignty. This agreement, described as "groundbreaking," is based on TTP's Māori data sovereignty framework, which has been under development for the past three years. 

As anchor tenants for Microsoft's data centres, TTP will play a pivotal role in safeguarding Māori data as a precious asset in an increasingly digital world. Ruth Russell, Te Tumu Paeroa’s Kaitautari Pārongo Matua (Chief Information Officer), emphasized the significance of protecting Māori data, describing it as a "taonga" or treasure. Anchor tenancy enables TTP to host data in Aotearoa, ensuring it remains within the country's sovereign borders. 

The agreement aims to deepen connections between landowners and their whenua (land) and facilitate faster recovery from major weather events while supporting innovation on key issues such as climate change. TTP's services include trust administration, property management, income distribution, and client fund management, making this partnership crucial for enhancing Māori data sovereignty. One of the primary benefits of the new cloud service is that data stored at the centre will not leave New Zealand's sovereign borders, ensuring compliance with local laws and regulations. 

This advanced data residency feature offered by Microsoft instills confidence that data resides in the desired territory, aligning with TTP's framework and recognizing the sovereignty of Māori data. Dan Te Whenua Walker from Microsoft highlights the opportunity for Māori to leverage artificial intelligence (AI) while acknowledging some uncertainties regarding its cultural implications. He emphasizes the importance of TTP's framework in guiding the adoption of AI, ensuring it aligns with Māori aspirations and values. DDS IT, responsible for migrating data to Microsoft's cloud servers, considers this partnership a unique opportunity. The data migration process involves transferring data between locations and formats, with the full transfer expected to take between 12 to 24 months. 

Moreover, the new data centre is set to be the most sustainable globally, emphasizing energy efficiency and environmental considerations. The partnership between TTP and Microsoft represents a significant step towards advancing Māori data sovereignty and leveraging technology to benefit indigenous communities. By hosting data within Aotearoa's sovereign borders and adhering to Māori principles of kaitiakitanga (guardianship), this collaboration sets a precedent for indigenous data governance worldwide.

Bridging the Gap Between Cloud vs On-premise Security

 

In the current landscape, the prevalence of the cloud era is undeniable, and the market is characterized by constant dynamism. Enterprises, in order to maintain relevance amid this competitive environment, are unmistakably demonstrating a keen interest in embracing cloud technologies. What motivates this significant shift? 

Cloud-centric security strategies, exemplified by initiatives like Secure Access Service Edge (SASE) and Security Service Edge (SSE), encompassing components such as Secure Web Gateway (SWG), Cloud Access Security Brokers (CASB), Data Loss Prevention (DLP), and Zero Trust Network Access (ZTNA), efficiently extend security to wherever corporate users, devices, and resources are located—leveraging the cloud as the central hub. 

With all security functionalities seamlessly delivered and managed through a unified interface, the security of both inbound and outbound traffic, often referred to as north-south traffic, is significantly fortified. 

On the flip side, the internal network's east-west traffic, which moves within the confines of data centers and the network but does not cross the network perimeter, remains untouched by the security checks implemented through cloud-based measures. 

A potential workaround involves keeping a traditional data center firewall dedicated to overseeing and regulating internal, east-west traffic. However, this hybrid security approach introduces increased expenses and intricacies in handling diverse security solutions. Many organizations strive to address these challenges by opting for integrated, cloud-based security stacks to streamline management and mitigate the complexities associated with maintaining separate security measures. 

To ensure comprehensive security coverage for organizations, a solution is required that safeguards both north-south and east-west traffic. The key lies in orchestration through a centralized, cloud-based console. Achieving this can be approached in two ways: 

1. Via WAN Firewall Policy 

Cloud-native security frameworks like SASE and SSE can provide east-west protection by directing internal traffic through the nearest point of presence (PoP). Unlike traditional local firewalls with their own setup limitations, SSE PoP allows firewall policies to be managed centrally through the platform's console. Admins can easily create access rules in the unified console, such as permitting authorized users on the corporate VLAN with approved, Active Directory-registered devices to access specific resources in the on-premise data center, following Zero Trust Network Access (ZTNA) principles. 

2. Via LAN Firewall Policy 

In a security-conscious scenario, where an IoT VLAN's CCTV camera needs access to an internal server, disabling default internet/WAN access is wise to prevent cyber threats. Implementing data center firewall policies at the Point of Presence (PoP) may not affect devices like IoT cameras with no internet access. 

SASE and SSE platforms address this by empowering administrators to set firewall policies on the local SD-WAN device. Organizations connect to SASE/SSE PoPs through this SD-WAN device, allowing direct rule configuration for internal LAN traffic. Pre-defined LAN firewall policies are locally enforced, with unmatched traffic sent to the PoP for further assessment, enhancing security management efficiency.

Vietnamese Cybercriminals Exploit Malvertising to Target Facebook Business Accounts

Cybercriminals associated with the Vietnamese cybercrime ecosystem are exploiting social media platforms, including Meta-owned Facebook, as a means to distribute malware. 

According to Mohammad Kazem Hassan Nejad, a researcher from WithSecure, malicious actors have been utilizing deceptive ads to target victims with various scams and malvertising schemes. This tactic has become even more lucrative with businesses increasingly using social media for advertising, providing attackers with a new type of attack vector – hijacking business accounts.

Over the past year, cyber attacks against Meta Business and Facebook accounts have gained popularity, primarily driven by activity clusters like Ducktail and NodeStealer, known for targeting businesses and individuals operating on Facebook. 

Social engineering plays a crucial role in gaining unauthorized access to user accounts, with victims being approached through platforms such as Facebook, LinkedIn, WhatsApp, and freelance job portals like Upwork. Search engine poisoning is another method employed to promote fake software, including CapCut, Notepad++, OpenAI ChatGPT, Google Bard, and Meta Threads.

Common tactics among these cybercrime groups include the misuse of URL shorteners, the use of Telegram for command-and-control (C2), and legitimate cloud services like Trello, Discord, Dropbox, iCloud, OneDrive, and Mediafire to host malicious payloads.

Ducktail, for instance, employs lures related to branding and marketing projects to infiltrate individuals and businesses on Meta's Business platform. In recent attacks, job and recruitment-related themes have been used to activate infections. 

Potential targets are directed to fraudulent job postings on platforms like Upwork and Freelancer through Facebook ads or LinkedIn InMail. These postings contain links to compromised job description files hosted on cloud storage providers, leading to the deployment of the Ducktail stealer malware.

The Ducktail malware is designed to steal saved session cookies from browsers, with specific code tailored to take over Facebook business accounts. These compromised accounts are sold on underground marketplaces, fetching prices ranging from $15 to $340.

Recent attack sequences observed between February and March 2023 involve the use of shortcut and PowerShell files to download and launch the final malware. The malware has evolved to harvest personal information from various platforms, including X (formerly Twitter), TikTok Business, and Google Ads. It also uses stolen Facebook session cookies to create fraudulent ads and gain elevated privileges.

One of the primary methods used to take over a victim's compromised account involves adding the attacker's email address, changing the password, and locking the victim out of their Facebook account.

The malware has incorporated new features, such as using RestartManager (RM) to kill processes that lock browser databases, a technique commonly found in ransomware. Additionally, the final payload is obfuscated using a loader to dynamically decrypt and execute it, making analysis and detection more challenging.

To hinder analysis efforts, the threat actors use uniquely generated assembly names and rely on SmartAssembly, bloating, and compression to obfuscate the malware.

Researchers from Zscaler also observed instances where the threat actors initiated contact using compromised LinkedIn accounts belonging to users in the digital marketing field, leveraging the authenticity of these accounts to aid in social engineering tactics. This highlights the worm-like propagation of Ducktail, where stolen LinkedIn credentials and cookies are used to log in to victims' accounts and expand their reach.

Ducktail is just one of many Vietnamese threat actors employing shared tools and tactics for fraudulent schemes. A Ducktail copycat known as Duckport, which emerged in late March 2023, engages in information stealing and Meta Business account hijacking. Notably, Duckport differs from Ducktail in terms of Telegram channels used for command and control, source code implementation, and distribution, making them distinct threats.

Duckport employs a unique technique of sending victims links to branded sites related to the impersonated brand or company, redirecting them to download malicious archives from file hosting services. Unlike Ducktail, Duckport replaces Telegram as a channel for passing commands to victims' machines and incorporates additional information stealing and account hijacking capabilities, along with taking screenshots and abusing online note-taking services as part of its command and control chain.

"The Vietnamese-centric element of these threats and high degree of overlaps in terms of capabilities, infrastructure, and victimology suggests active working relationships between various threat actors, shared tooling and TTPs across these threat groups, or a fractured and service-oriented Vietnamese cybercriminal ecosystem (akin to ransomware-as-a-service model) centered around social media platforms such as Facebook," WithSecure said.