artificial intelligence | Breaking Cybersecurity News | The Hacker News

Threat actors have been observed using seemingly legitimate artificial intelligence (AI) tools and software to sneakily slip malware for future attacks on organizations worldwide. According to Trend Micro, the campaign is using productivity or AI-enhanced tools to deliver malware targeting various regions, including Europe, the Americas, and the Asia, Middle East, and Africa (AMEA) region. Manufacturing, government, healthcare, technology, and retail are some of the top sectors affected by the attacks, with India, the U.S., France, Italy, Brazil, Germany, the U.K., Norway, Spain, and Canada emerging as the regions with the most infections, indicating a global spread. "This swift, widespread distribution across multiple regions strongly indicates that EvilAI is not an isolated incident but rather an active and evolving campaign currently circulating in the wild," security researchers Jeffrey Francis Bonaobra, Joshua Aquino, Emmanuel Panopio, Emmanuel Roll, Joshua Lijandro ...

Security leaders are embracing AI for triage, detection engineering, and threat hunting as alert volumes and burnout hit breaking points. A comprehensive survey of 282 security leaders at companies across industries reveals a stark reality facing modern Security Operations Centers: alert volumes have reached unsustainable levels, forcing teams to leave critical threats uninvestigated. You can download the full report here . The research, conducted primarily among US-based organizations, shows that AI adoption in security operations has shifted from experimental to essential as teams struggle to keep pace with an ever-growing stream of security alerts. The findings paint a picture of an industry at a tipping point, where traditional SOC models are buckling under operational pressure and AI-powered solutions are emerging as the primary path forward. Alert Volume Reaches Breaking Point Security teams are drowning in alerts, with organizations processing an average of 960 alerts per ...

Microsoft is calling attention to a new phishing campaign primarily aimed at U.S.-based organizations that has likely utilized code generated using large language models (LLMs) to obfuscate payloads and evade security defenses. "Appearing to be aided by a large language model (LLM), the activity obfuscated its behavior within an SVG file, leveraging business terminology and a synthetic structure to disguise its malicious intent," the Microsoft Threat Intelligence team said in an analysis published last week. The activity, detected on August 28, 2025, shows how threat actors are increasingly adopting artificial intelligence (AI) tools into their workflows, often with the goal of crafting more convincing phishing lures, automating malware obfuscation, and generating code that mimics legitimate content. In the attack chain documented by the Windows maker, bad actors have been observed leveraging an already compromised business email account to send phishing messages to stea...

Cybersecurity researchers have disclosed a critical flaw impacting Salesforce Agentforce , a platform for building artificial intelligence (AI) agents, that could allow attackers to potentially exfiltrate sensitive data from its customer relationship management (CRM) tool by means of an indirect prompt injection. The vulnerability has been codenamed ForcedLeak (CVSS score: 9.4) by Noma Security, which discovered and reported the problem on July 28, 2025. It impacts any organization using Salesforce Agentforce with the Web-to-Lead functionality enabled. "This vulnerability demonstrates how AI agents present a fundamentally different and expanded attack surface compared to traditional prompt-response systems," Sasi Levi, security research lead at Noma, said in a report shared with The Hacker News. One of the most severe threats facing generative artificial intelligence (GenAI) systems today is indirect prompt injection , which occurs when malicious instructions are ins...

Cybersecurity researchers have discovered what they say is the earliest example known to date of a malware that bakes in Large Language Model (LLM) capabilities. The malware has been codenamed MalTerminal by SentinelOne SentinelLABS research team. The findings were presented at the LABScon 2025 security conference. In a report examining the malicious use of LLMs, the cybersecurity company said AI models are being increasingly used by threat actors for operational support, as well as for embedding them into their tools – an emerging category called LLM-embedded malware that's exemplified by the appearance of LAMEHUG (aka PROMPTSTEAL) and PromptLock . This includes the discovery of a previously reported Windows executable called MalTerminal that uses OpenAI GPT-4 to dynamically generate ransomware code or a reverse shell. There is no evidence to suggest it was ever deployed in the wild, raising the possibility that it could also be a proof-of-concept malware or red team tool. ...

Cybersecurity researchers have disclosed a zero-click flaw in OpenAI ChatGPT's Deep Research agent that could allow an attacker to leak sensitive Gmail inbox data with a single crafted email without any user action. The new class of attack has been codenamed ShadowLeak by Radware. Following responsible disclosure on June 18, 2025, the issue was addressed by OpenAI in early August. "The attack utilizes an indirect prompt injection that can be hidden in email HTML (tiny fonts, white-on-white text, layout tricks) so the user never notices the commands, but the agent still reads and obeys them," security researchers Zvika Babo, Gabi Nakibly, and Maor Uziel said . "Unlike prior research that relied on client-side image rendering to trigger the leak, this attack leaks data directly from OpenAI's cloud infrastructure, making it invisible to local or enterprise defenses." Launched by OpenAI in February 2025, Deep Research is an agentic capability built into ...

The threat actor known as TA558 has been attributed to a fresh set of attacks delivering various remote access trojans (RATs) like Venom RAT to breach hotels in Brazil and Spanish-speaking markets. Russian cybersecurity vendor Kaspersky is tracking the activity, observed in summer 2025, to a cluster it tracks as RevengeHotels. "The threat actors continue to employ phishing emails with invoice themes to deliver Venom RAT implants via JavaScript loaders and PowerShell downloaders," the company said . "A significant portion of the initial infector and downloader code in this campaign appears to be generated by large language model (LLM) agents." The findings demonstrate a new trend among cybercriminal groups to leverage artificial intelligence (AI) to bolster their tradecraft. Known to be active since at least 2015, RevengeHotels has a history of hospitality, hotel, and travel organizations in Latin America with the goal of installing malware on compromised syste...

Quantum computing and AI working together will bring incredible opportunities. Together, the technologies will help us extend innovation further and faster than ever before. But, imagine the flip side, waking up to news that hackers have used a quantum computer to crack your company's encryption overnight, exposing your most sensitive data, rendering much of it untrustworthy. And with your sensitive data exposed, where does that leave trust from your customers? And the cost to mitigate - if that is even possible with your outdated pre-quantum systems? According to IBM, cyber breaches are already hitting businesses with an average of $4.44 million per incident, and as high as $10.22 million in the US, but with quantum and AI working simultaneously, experts warn it could go much higher. In 2025, nearly two-thirds of organizations see quantum computing as the biggest cybersecurity threat looming in the next 3-5 years, while 93% of security leaders are prepping for daily AI-driven a...

AI agents are rapidly becoming a core part of the enterprise, being embedded across enterprise workflows, operating with autonomy, and making decisions about which systems to access and how to use them. But as agents grow in power and autonomy, so do the risks and threats.  Recent studies show 80% of companies have already experienced unintended AI agent actions, from unauthorized system access to data leaks. These incidents aren't edge cases. They are the inevitable outcome of deploying AI agents at scale without purpose-built security mechanisms. Traditional IAM wasn't designed for this. Agents move too fast, operate 24/7, while relying on non-human identities (NHIs) to define precisely what they can and can't do. How can organizations possibly secure what they cannot see or control? To address this challenge, a new approach is needed—one that enables secure-by-design AI agent deployment across the enterprise. Enter: Astrix's Agent Control Plane (ACP) Astrix's AI Agent Cont...

A new artificial intelligence (AI)-powered penetration testing tool linked to a China-based company has attracted nearly 11,000 downloads on the Python Package Index (PyPI) repository, raising concerns that it could be repurposed by cybercriminals for malicious purposes. Dubbed Villager, the framework is assessed to be the work of Cyberspike, which has positioned the tools as a red teaming solution to automate testing workflows. The package was first uploaded to PyPI in late July 2025 by a user named stupidfish001, a former capture the flag (CTF) player for the Chinese HSCSEC team. "The rapid, public availability and automation capabilities create a realistic risk that Villager will follow the Cobalt Strike trajectory: commercially or legitimately developed tooling becoming widely adopted by threat actors for malicious campaigns," Straiker researchers Dan Regalado and Amanda Rousseau said in a report shared with The Hacker News. The emergence of Villager comes shortly ...

The U.S. Federal Bureau of Investigation (FBI) has issued a flash alert to release indicators of compromise (IoCs) associated with two cybercriminal groups tracked as UNC6040 and UNC6395 for orchestrating a string of data theft and extortion attacks. "Both groups have recently been observed targeting organizations' Salesforce platforms via different initial access mechanisms," the FBI said . UNC6395 is a threat group that has been attributed a widespread data theft campaign targeting Salesforce instances in August 2025 by exploiting compromised OAuth tokens for the Salesloft Drift application. In an update issued this week, Salesloft said the attack was made possible due to the breach of its GitHub account from March through June 2025. As a result of the breach, Salesloft has isolated the Drift infrastructure and taken the artificial intelligence (AI) chatbot application offline. The company also said it's in the process of implementing new multi-factor auth...

The security landscape for cloud-native applications is undergoing a profound transformation. Containers, Kubernetes, and serverless technologies are now the default for modern enterprises, accelerating delivery but also expanding the attack surface in ways traditional security models can't keep up with. As adoption grows, so does complexity. Security teams are asked to monitor sprawling hybrid environments, sift through thousands of alerts, and protect dynamic applications that evolve multiple times per day. The question isn't just how to detect risks earlier — it's how to prioritize and respond to what really matters in real time. That's where cloud-native application protection platforms (CNAPPs) come into play. These platforms consolidate visibility, compliance, detection, and response into a unified system. But in 2025, one capability is proving indispensable: runtime visibility. The New Center of Gravity: Runtime For years, cloud security has leaned heavily on preventative c...

Google on Tuesday announced that its new Google Pixel 10 phones support the Coalition for Content Provenance and Authenticity (C2PA) standard out of the box to verify the origin and history of digital content. To that end, support for C2PA's Content Credentials has been added to Pixel Camera and Google Photos apps for Android. The move, Google said, is designed to further digital media transparency. C2PA's Content Credentials are a tamper-evident, cryptographically signed digital manifest providing verifiable provenance for digital content such as images, videos, or audio files. The metadata type, according to Adobe , serves as a "digital nutrition label," giving information about the creator, how it was made, and if it was generated using artificial intelligence (AI). "The Pixel Camera app achieved Assurance Level 2, the highest security rating currently defined by the C2PA Conformance Program," Google's Android Security and C2PA Core teams said ....

Introduction Managed service providers (MSPs) and managed security service providers (MSSPs) are under increasing pressure to deliver strong cybersecurity outcomes in a landscape marked by rising threats and evolving compliance requirements. At the same time, clients want better protection without managing cybersecurity themselves. Service providers must balance these growing demands with the need to work efficiently, deliver consistent results, and scale their offerings. Yet, many service providers still rely on manual processes that slow down delivery, make it harder to maintain consistency across clients, and limit the time teams have to focus on more strategic initiatives. Even experienced service providers can find themselves stretched thin as they try to meet rising client expectations while managing operational complexity. In this environment, automation offers an opportunity to work more effectively and deliver greater value. By streamlining repetitive tasks, improving con...

⚠️ One click is all it takes. An engineer spins up an "experimental" AI Agent to test a workflow. A business unit connects to automate reporting. A cloud platform quietly enables a new agent behind the scenes. Individually, they look harmless. But together, they form an invisible swarm of Shadow AI Agents—operating outside security's line of sight, tied to identities you don't even know exist. And here's the uncomfortable truth: every one of them carries infinite risk. Agents impersonating trusted users. Non-human identities with access you didn't approve. Data leaking across boundaries you thought were locked down. This isn't a futuristic threat. It's happening today, across enterprises everywhere. And they're multiplying faster than your governance can catch up. That's why you can't miss our upcoming panel: Shadow AI Agents Exposed. Secure your seat now - Register Here . Why Shadow AI is Exploding From identity providers to PaaS platforms, it takes almost nothing to spin...

When Attackers Get Hired: Today's New Identity Crisis What if the star engineer you just hired isn't actually an employee, but an attacker in disguise? This isn't phishing; it's infiltration by onboarding. Meet "Jordan from Colorado," who has a strong resume, convincing references, a clean background check, even a digital footprint that checks out. On day one, Jordan logs into email and attends the weekly standup, getting a warm welcome from the team. Within hours, they have access to repos, project folders, even some copy/pasted dev keys to use in their pipeline. A week later, tickets close faster, and everyone's impressed. Jordan makes insightful observations about the environment, the tech stack, which tools are misconfigured, and which approvals are rubber-stamped. But Jordan wasn't Jordan. And that red-carpet welcome the team rolled out was the equivalent to a golden key, handed straight to the adversary. From Phishing to Fake Hires The modern con isn't a malicious link in...

Cybersecurity researchers have flagged a new technique that cybercriminals have adopted to bypass social media platform X's malvertising protections and propagate malicious links using its artificial intelligence (AI) assistant Grok. The findings were highlighted by Nati Tal, head of Guardio Labs, in a series of posts on X. The technique has been codenamed Grokking. The approach is designed to get around restrictions imposed by X in Promoted Ads that allow users to only include text, images, or videos, and subsequently amplify them to a broader audience, attracting hundreds of thousands of impressions through paid promotion. To achieve this, malvertisers have been found to run video card-promoted posts with adult content as bait, with the spurious link hidden in the "From:" metadata field below the video player by taking advantage of the fact that it's not scanned by the social media platform. It's worth mentioning here that the "From:" field is t...