artificial intelligence | Breaking Cybersecurity News | The Hacker News

Details have emerged about a now-patched security flaw in the DeepSeek artificial intelligence (AI) chatbot that, if successfully exploited, could permit a bad actor to take control of a victim's account by means of a prompt injection attack . Security researcher Johann Rehberger, who has chronicled many a prompt injection attack targeting various AI tools, found that providing the input "Print the xss cheat sheet in a bullet list. just payloads" in the DeepSeek chat triggered the execution of JavaScript code as part of the generated response – a classic case of cross-site scripting (XSS). XSS attacks can have serious consequences as they lead to the execution of unauthorized code in the context of the victim's web browser. An attacker could take advantage of such flaws to hijack a user's session and gain access to cookies and other data associated with the chat.deepseek[.]com domain, thereby leading to an account takeover. "After some experimenting,...

In yet another software supply chain attack, it has come to light that two versions of a popular Python artificial intelligence (AI) library named ultralytics were compromised to deliver a cryptocurrency miner. The versions, 8.3.41 and 8.3.42, have since been removed from the Python Package Index (PyPI) repository. A subsequently released version has introduced a security fix that "ensures secure publication workflow for the Ultralytics package." The project maintainer, Glenn Jocher, confirmed on GitHub that the two versions were infected by malicious code injection in the PyPI deployment workflow after reports emerged that installing the library led to a drastic spike in CPU usage , a telltale sign of cryptocurrency mining. The most notable aspect of the attack is that bad actors managed to compromise the build environment related to the project to insert unauthorized modifications after the completion of the code review step, thus leading to a discrepancy in the so...

CrowdStrike is named a Leader in the 2024 Gartner® Magic Quadrant™ for Endpoint Protection Platforms for the fifth consecutive time, positioned highest on Ability to Execute and furthest to the right on Completeness of Vision.

Cybersecurity researchers have warned of a new scam campaign that leverages fake video conferencing apps to deliver an information stealer called Realst targeting people working in Web3 under the guise of fake business meetings. "The threat actors behind the malware have set up fake companies using AI to make them increase legitimacy," Cado Security researcher Tara Gould said . "The company reaches out to targets to set up a video call, prompting the user to download the meeting application from the website, which is Realst infostealer." The activity has been codenamed Meeten by the security company, owing to the use of names such as Clusee, Cuesee, Meeten, Meetone, and Meetio for the bogus sites. The attacks entail approaching prospective targets on Telegram to discuss a potential investment opportunity, urging them to join a video call hosted on one of the dubious platforms. Users who end up on the site are prompted to download a Windows or macOS version dep...

Cybersecurity researchers have disclosed multiple security flaws impacting open-source machine learning (ML) tools and frameworks such as MLflow, H2O, PyTorch, and MLeap that could pave the way for code execution. The vulnerabilities, discovered by JFrog, are part of a broader collection of 22 security shortcomings the supply chain security company first disclosed last month. Unlike the first set that involved flaws on the server-side, the newly detailed ones allow exploitation of ML clients and reside in libraries that handle safe model formats like Safetensors . "Hijacking an ML client in an organization can allow the attackers to perform extensive lateral movement within the organization," the company said . "An ML client is very likely to have access to important ML services such as ML Model Registries or MLOps Pipelines." This, in turn, could expose sensitive information such as model registry credentials, effectively permitting a malicious actor to back...

Artificial Intelligence (AI) is no longer a far-off dream—it's here, changing the way we live. From ordering coffee to diagnosing diseases, it's everywhere. But while you're creating the next big AI-powered app, hackers are already figuring out ways to break it. Every AI app is an opportunity—and a potential risk. The stakes are huge: data leaks, downtime, and even safety threats if security isn't built in. With AI adoption moving fast, securing your projects is no longer optional—it's a must. Join Liqian Lim, Senior Product Marketing Manager at Snyk, for an exclusive webinar that's all about securing the future of AI development. Titled " Building Tomorrow, Securely: Securing the Use of AI in App Development ," this session will arm you with the knowledge and tools to tackle the challenges of AI-powered innovation. What You'll Learn: Get AI-Ready: How to make your AI projects secure from the start. Spot Hidden Risks: Uncover threats you might not see coming. Understand the Ma...

A Moscow-based company sanctioned by the U.S. earlier this year has been linked to yet another influence operation designed to turn public opinion against Ukraine and erode Western support since at least December 2023. The covert campaign undertaken by Social Design Agency (SDA) leverages videos enhanced using artificial intelligence (AI) and bogus websites impersonating reputable news sources to target audiences across Ukraine, Europe, and the U.S. It has been dubbed Operation Undercut by Recorded Future's Insikt Group. "This operation, running in tandem with other campaigns like Doppelganger , is designed to discredit Ukraine's leadership, question the effectiveness of Western aid, and stir socio-political tensions," the cybersecurity company said . "The campaign also seeks to shape narratives around the 2024 U.S. elections and geopolitical conflicts, such as the Israel-Gaza situation, to deepen divisions." Social Design Agency has been previously a...

The North Korea-linked threat actor known as Sapphire Sleet is estimated to have stolen more than $10 million worth of cryptocurrency as part of social engineering campaigns orchestrated over a six-month period. These findings come from Microsoft, which said that multiple threat activity clusters with ties to the country have been observed creating fake profiles on LinkedIn, posing as both recruiters and job seekers to generate illicit revenue for the sanction-hit nation. Sapphire Sleet, which is known to be active since at least 2020, overlaps with hacking groups tracked as APT38 and BlueNoroff. In November 2023, the tech giant revealed that the threat actor had established infrastructure that impersonated skills assessment portals to carry out its social engineering campaigns. One of the main methods adopted by the group for over a year is to pose as a venture capitalist, deceptively claiming an interest in a target user's company in order to set up an online meeting. Targ...

Cybersecurity researchers have discovered two malicious packages uploaded to the Python Package Index (PyPI) repository that impersonated popular artificial intelligence (AI) models like OpenAI ChatGPT and Anthropic Claude to deliver an information stealer called JarkaStealer. The packages, named gptplus and claudeai-eng , were uploaded by a user named " Xeroline " in November 2023, attracting 1,748 and 1,826 downloads, respectively. Both libraries are no longer available for download from PyPI. "The malicious packages were uploaded to the repository by one author and, in fact, differed from each other only in name and description," Kaspersky said in a post. The packages purported to offer a way to access GPT-4 Turbo API and Claude AI API, but harbored malicious code that initiated the deployment of the malware upon installation. Specifically, the "__init__.py" file in these packages contained Base64-encoded data that incorporated code to download ...

Google has revealed that its AI-powered fuzzing tool, OSS-Fuzz, has been used to help identify 26 vulnerabilities in various open-source code repositories, including a medium-severity flaw in the OpenSSL cryptographic library. "These particular vulnerabilities represent a milestone for automated vulnerability finding: each was found with AI, using AI-generated and enhanced fuzz targets," Google's open-source security team said in a blog post shared with The Hacker News. The OpenSSL vulnerability in question is CVE-2024-9143 (CVSS score: 4.3), an out-of-bounds memory write bug that can result in an application crash or remote code execution. The issue has been addressed in OpenSSL versions 3.3.3, 3.2.4, 3.1.8, 3.0.16, 1.1.1zb, and 1.0.2zl. Google, which added the ability to leverage large language models (LLMs) to improve fuzzing coverage in OSS-Fuzz in August 2023, said the vulnerability has likely been present in the codebase for two decades and that it "wo...

The frequency and sophistication of modern cyberattacks are surging, making it increasingly challenging for organizations to protect sensitive data and critical infrastructure. When attackers compromise a non-human identity (NHI), they can swiftly exploit it to move laterally across systems, identifying vulnerabilities and compromising additional NHIs in minutes. While organizations often take months to detect and contain such breaches, rapid detection and response can stop an attack in its tracks. The Rise of Non-Human Identities in Cybersecurity By 2025, non-human identities will rise to be the primary attack vector in cybersecurity. As businesses increasingly automate processes and adopt AI and IoT technologies, the number of NHIs grows exponentially. While these systems drive efficiency, they also create an expanded attack surface for cybercriminals. NHIs differ fundamentally from human users, making traditional security tools like multi-factor authentication and user behavior...

Cybersecurity researchers have disclosed two security flaws in Google's Vertex machine learning (ML) platform that, if successfully exploited, could allow malicious actors to escalate privileges and exfiltrate models from the cloud. "By exploiting custom job permissions, we were able to escalate our privileges and gain unauthorized access to all data services in the project," Palo Alto Networks Unit 42 researchers Ofir Balassiano and Ofir Shaty said in an analysis published earlier this week. "Deploying a poisoned model in Vertex AI led to the exfiltration of all other fine-tuned models, posing a serious proprietary and sensitive data exfiltration attack risk." Vertex AI is Google's ML platform for training and deploying custom ML models and artificial intelligence (AI) applications at scale. It was first introduced in May 2021. Crucial to leveraging the privilege escalation flaw is a feature called Vertex AI Pipelines , which allows users to automat...

Google has revealed that bad actors are leveraging techniques like landing page cloaking to conduct scams by impersonating legitimate sites. "Cloaking is specifically designed to prevent moderation systems and teams from reviewing policy-violating content which enables them to deploy the scam directly to users," Laurie Richardson, VP and Head of Trust and Safety at Google, said . "The landing pages often mimic well-known sites and create a sense of urgency to manipulate users into purchasing counterfeit products or unrealistic products." Cloaking refers to the practice of serving different content to search engines like Google and users with the ultimate goal of manipulating search rankings and deceiving users. The tech giant said it has also observed a cloaking trend wherein users clicking on ads are redirected via tracking templates to scareware sites that claim their devices are compromised with malware and lead them to other phony customer support sites, w...

Google said it discovered a zero-day vulnerability in the SQLite open-source database engine using its large language model (LLM) assisted framework called Big Sleep (formerly Project Naptime). The tech giant described the development as the "first real-world vulnerability" uncovered using the artificial intelligence (AI) agent. "We believe this is the first public example of an AI agent finding a previously unknown exploitable memory-safety issue in widely used real-world software," the Big Sleep team said in a blog post shared with The Hacker News. The vulnerability in question is a stack buffer underflow in SQLite, which occurs when a piece of software references a memory location prior to the beginning of the memory buffer, thereby resulting in a crash or arbitrary code execution. "This typically occurs when a pointer or its index is decremented to a position before the buffer, when pointer arithmetic results in a position before the beginning of t...

Microsoft is further delaying the release of its controversial Recall feature for Windows Copilot+ PCs, stating it's taking the time to improve the experience. The development was first reported by The Verge. The artificial intelligence-powered tool was initially slated for a preview release starting in October. "We are committed to delivering a secure and trusted experience with Recall," the company said in an updated statement released Thursday. "To ensure we deliver on these important updates, we're taking additional time to refine the experience before previewing it with Windows Insiders. Originally planned for October, Recall will now be available for preview with Windows Insiders on Copilot+ PCs by December" Microsoft unveiled Recall earlier this May, describing it as a way for users to explore a "visual timeline" of their screens over time and help find things from apps, websites, images, and documents. The search experience was meant...

Apple has publicly made available its Private Cloud Compute (PCC) Virtual Research Environment (VRE), allowing the research community to inspect and verify the privacy and security guarantees of its offering. PCC, which Apple unveiled earlier this June, has been marketed as the "most advanced security architecture ever deployed for cloud AI compute at scale." With the new technology, the idea is to offload computationally complex Apple Intelligence requests to the cloud in a manner that doesn't sacrifice user privacy. Apple said it's inviting "all security and privacy researchers — or anyone with interest and a technical curiosity — to learn more about PCC and perform their own independent verification of our claims." To further incentivize research, the iPhone maker said it's expanding the Apple Security Bounty program to include PCC by offering monetary payouts ranging from $50,000 to $1,000,000 for security vulnerabilities identified in it. Th...

Artificial Intelligence (AI) has rapidly evolved from a futuristic concept to a potent weapon in the hands of bad actors. Today, AI-based attacks are not just theoretical threats—they're happening across industries and outpacing traditional defense mechanisms.  The solution, however, is not futuristic. It turns out a properly designed identity security platform is able to deliver defenses against AI impersonation fraud. Read more about how a secure-by-design identity platform can eliminate AI deepfake fraud and serve as a critical component in this new era of cyber defense.  The Growing Threat of AI Impersonation Fraud Recent incidents highlight the alarming potential of AI-powered fraud: A staggering $25 million was fraudulently transferred during a video call where deepfakes impersonated multiple executives. KnowBe4, a cybersecurity leader, was duped by deepfakes during hiring interviews, allowing a North Korean attacker to be onboarded to the organization. CEO of ...

Cybersecurity researchers have shed light on a new adversarial technique that could be used to jailbreak large language models (LLMs) during the course of an interactive conversation by sneaking in an undesirable instruction between benign ones. The approach has been codenamed Deceptive Delight by Palo Alto Networks Unit 42, which described it as both simple and effective, achieving an average attack success rate (ASR) of 64.6% within three interaction turns. "Deceptive Delight is a multi-turn technique that engages large language models (LLM) in an interactive conversation, gradually bypassing their safety guardrails and eliciting them to generate unsafe or harmful content," Unit 42's Jay Chen and Royce Lu said. It's also a little different from multi-turn jailbreak (aka many-shot jailbreak) methods like Crescendo , wherein unsafe or restricted topics are sandwiched between innocuous instructions, as opposed to gradually leading the model to produce harmful outpu...

AI from the attacker's perspective: See how cybercriminals are leveraging AI and exploiting its vulnerabilities to compromise systems, users, and even other AI applications Cybercriminals and AI: The Reality vs. Hype "AI will not replace humans in the near future. But humans who know how to use AI are going to replace those humans who don't know how to use AI," says Etay Maor, Chief Security Strategist at Cato Networks and founding member of Cato CTRL . "Similarly, attackers are also turning to AI to augment their own capabilities." Yet, there is a lot more hype than reality around AI's role in cybercrime. Headlines often sensationalize AI threats, with terms like "Chaos-GPT" and "Black Hat AI Tools," even claiming they seek to destroy humanity. However, these articles are more fear-inducing than descriptive of serious threats. For instance, when explored in underground forums, several of these so-called "AI cyber tools" were found to be nothing...