artificial intelligence | Breaking Cybersecurity News | The Hacker News

Google's DeepMind on Tuesday announced the release of Gemini 3.5 Flash Cyber , a specialized artificial intelligence (AI) model built atop 3.5 Flash that's designed to discover, validate, and patch vulnerabilities quickly and efficiently. According to the tech giant, the model will be exclusively available to governments and trusted partners via CodeMender as part of a limited-access pilot program. CodeMender is an AI-powered agent for vulnerability discovery and patching that was unveiled by the company in October 2025. A Google DeepMind spokesperson told The Hacker News that there are plans to extend the model's capabilities to include red-teaming features and end-to-end enterprise defense. The lightweight model, per DeepMind, is both cost-efficient and highly capable alternative to large, costly cybersecurity-focused models. CodeMender can call upon 3.5 Flash Cyber "multiple times at high speed and low cost," allowing the AI agent to scan more code path...

Every patch is a confession. The moment a vendor ships a security fix, the diff between the old code and the new code tells anyone watching exactly what was broken and where. Turn that diff back into a working exploit, and you can hit every system that hasn't updated yet. This is N-day exploitation, and it's always been a race: the vendor patches, the clock starts, and defenders try to deploy before an attacker finishes reverse-engineering the fix. For the last thirty-odd years, defenders usually won that race. Reverse-engineering a patch into a reliable exploit was slow, specialized work - usually requiring weeks of expert-level effort. Historically, the gap between a patch and a working public exploit spanned weeks, often months. The traditional playbook assumed you had at least a few weeks. You don't anymore. Not even close. Reverse-engineering a patch used to take weeks. Mythos does it in an hour. Anthropic's red team measured exactly that. Given not...

Researchers at Sysdig have linked a second attack on the same Langflow server to JADEPUFFER , the AI-agent-driven operator it first documented earlier this month. The same operator has now been spotted deploying ENCFORGE , a new compiled Go ransomware designed to encrypt model weights, vector indexes, training datasets, and other AI infrastructure files across the host filesystem. The entry point did not change. Langflow versions before 1.3.0 expose the /api/v1/validate/code endpoint without authentication, allowing any remote attacker to execute arbitrary Python on the server. The flaw, CVE-2025-3248 , carries a CVSS score of 9.8 and has been in CISA's Known Exploited Vulnerabilities catalog since May 5, 2025. As The Hacker News reported earlier this month, the prior operation used throwaway Python code and MySQL's AES_ENCRYPT() function to encrypt and destroy data in Nacos (Alibaba's configuration server) and production databases. The new ENCFORGE payload re...

Threat actors are now exploiting a recently disclosed critical security flaw impacting ServiceNow AI Platform, according to Defused Cyber . In a post shared on X, the threat intelligence firm said it's observing in-the-wild exploitation of CVE-2026-6875 (CVSS score: 9.5), a sandbox escape vulnerability that could allow an unauthenticated user to run arbitrary code. Patches for the flaw were released by ServiceNow throughout June in the following versions - Brazil EA and Brazil GA Australia Patch 2 Zurich Patch 7b and Zurich Patch 9 Yokohama Patch 12 Hot Fix 1b and Yokohama Patch 13 Searchlight Cyber, which disclosed additional technical specifics, said it reported the issue on April 1, 2026, adding it allows a complete compromise of the ServiceNow instance as well as all connected proxy servers. Besides rolling out a fix, ServiceNow is "enhancing instance security by severely restricting the type of code that can run in sandbox contexts," securit...

Cybersecurity researchers have discovered nearly 7,600 malicious GitHub repositories, out of which more than 800 pose as artificial intelligence (AI) skills or Model Context Protocol (MCP) servers to deliver a malware family known as SmartLoader as part of an ongoing campaign codenamed FakeGit . "FakeGit uses copied projects, lookalike developer profiles, convincing READMEs, and malicious ZIP files to deliver SmartLoader malware," Oleg Zaytsev, lead security researcher at Island, said in a report shared with The Hacker News. The end goal of these attacks is to leverage the access afforded by SmartLoader to establish persistence and push secondary payloads, such as StealC, an information stealer capable of harvesting a wide range of data from compromised systems. It's worth mentioning here that the use of trojanized MCP servers to distribute SmartLoader and StealC was flagged earlier this year by Straiker AI and subsequently by Derp.ca . But a concerning aspect ...

A malware operator left its delivery server wide open, and Rapid7 pulled down the whole toolkit: 1,048 files spanning lure templates, filename-spoofing tests, execution experiments, droppers, builder notes, and two campaign chains. One was already live against Windows users in Mexico, delivering an infostealer through a fake government ID-lookup site over WebDAV. What makes it more than a payload dump: it caught the operation mid-build. Testing notes, failed experiments, documentation, and live delivery logs sat in one place, the kind of complete development trail defenders rarely see. Rapid7 reads the artifacts, down to a hardcoded path pointing at an open-source AI coding tool, as an operator using generative AI to produce, test, and document phishing delivery at speed. The most developed test set focused on CVE-2025-33053 (CVSS 8.8, now in CISA's KEV catalog ), the WebDAV working-directory hijack Check Point documented last year in its Stealth Falcon reporting. The...

A single request should not be able to do this much. But this week, small inputs led to code execution, memory loss, stolen keys, and disabled security tools. The paths were often simple: exposed systems, weak checks, old drivers, fake prompts, and public code used for malware delivery. Some bugs were new. Others were already being used before defenders had time to patch. Here is the full recap of what broke, what was exploited, and what needs attention now. ⚡ Threat of the Week New wp2shell WordPress Core Flaw Lets Unauthenticated Attackers Run Code - Searchlight Cyber disclosed a pre-authenticated remote code execution vulnerability in WordPress Core that can be exploited anonymously on a standard WordPress installation, without requiring any plugins or other special conditions. It is a combination of CVE-2026-63030 (REST API batch-route confusion) and CVE-2026-60137 (SQL injection in WordPress core) that can be chained to turn an anonymous request into code execution. wa...

The industry spent the initial months after Anthropic's April 7 Mythos reveal focused on volume . How many new CVEs would Mythos add to an already overloaded pipeline? How quickly would the flood of AI-driven discovery overwhelm triage capabilities? How long would it take adversaries to weaponize Mythos findings at scale? Those questions were and remain valid. Yet they all stop short of addressing the single metric that determines whether any of those vulnerabilities actually lead to a breach: the exposure window. The exposure window - the gap between the moment a vulnerability becomes exploitable and the moment your team fixes it - is the time an attacker has to do actual damage. That window is currently open far too wide. In 2025, the average eCrime breakout time dropped to 29 minutes . Even PCI DSS - the strictest compliance framework in the industry - allows 30 days to remediate a critical vulnerability. That's a 1,000-to-1 gap between how fast attackers move and how f...

A solo Russian-speaking threat actor known as " bandcampro " outsourced a chunk of their operations to Google's open-source Gemini CLI artificial intelligence (AI) and commandeered a live botnet. The findings come from an analysis of 200 Gemini CLI session logs between March 19 and April 21, 2026, which found the threat actor using AI, among other things, to crack passwords, set up a residential proxy, compromise WordPress merchants, and plan a phone-based cryptocurrency fraud scheme aimed at elderly people in the U.S. and Canada. "The logs documented how the threat actor used an AI agent to migrate a command-and-control (C&C) server, and to control a small-scale botnet, among other hacking activities," Trend Micro researchers Joseph C Chen, Philippe Lin, Lucas Silva, Vladimir Kropotov, and Fyodor Yarochkin said . "The entire C&C operation fits in three plaintext files totaling roughly 5 KB, making it highly replicable and effectively disposa...

The European Commission on Thursday ordered Google to give rival AI assistants the same reach into Android that Gemini already has: the camera, the microphone, whatever is on screen, a wake word that fires with the display off, and the ability to drive other apps in the background by imitating taps and typing. Google has to ship it in the next major release, Android 18, and by 1 August 2027 at the latest. That is one of two binding specification decisions adopted on 16 July under the Digital Markets Act, six months after the Commission opened proceedings on 27 January. The second makes Google hand anonymised Search query, click, and ranking data to rival search engines, and to AI chatbots that do search, for a cost-based fee. Neither is a fine. Specification proceedings only say what a gatekeeper has to build; the Commission's separate power to open a non-compliance case , fines included, is untouched. Android carries around 60% of European mobile users. Five features g...

Military forces are under increasing pressure to field autonomous capabilities faster than ever before. Across the U.S., UK, and NATO, new investment, evolving defense strategies, and accelerated acquisition pathways are transforming how capability is delivered, rewarding programs that can move from concept to operational deployment at commercial speed. Now the focus shifts to the trusted information infrastructure that allows them to operate together at mission speed. As autonomous aircraft, uncrewed maritime vessels, ground systems, satellites and AI-enabled mission applications become increasingly connected, so too does the information that powers them. Telemetry, ISR, command data, AI outputs, sensor-to-shooter workflows and coalition intelligence all need to move seamlessly across platforms, domains, and partners. The future force won't be defined by autonomous systems alone, it will be defined by the trusted information infrastructure that connects them. Defense Has ...

A lot of this week’s trouble starts with something that looks close enough. A familiar repo. A useful installer. A harmless sync setting. Then the handoff goes bad, the box starts talking to someone else, and the damage moves faster than the explanation. Old bugs are back, weak defaults are earning their keep, and some attack paths are so plain they barely feel like research. Here’s the mess. Game cheats drop spyware 11 Malicious NuGet Tools Masquerade as Game Cheats to Drop Windows Surveillance Payload Cybersecurity researchers 11 malicious NuGet packages published as .NET command-line tools that present themselves as game utilities, bots, and "panels," each of which act as a first-stage downloader responsible for fetching and executing a second-stage Python payload named "pepesoft.exe" from GitHub Releases and Hugging Face paths under the username "pepegit666," along with a dormant BitTorrent fallback...

An advanced malware previously attributed to a China-linked threat actor has resurfaced after more than four years within a Taiwan manufacturing firm, along with a previously unreported backdoor dubbed Stupig . Daxin ("srt64.sys"), as the kernel-mode rootkit is referred to, was first documented by Broadcom-owned Symantec in March 2022, with evidence indicating its use in targeted attacks aimed at governments and other critical infrastructure targets since 2013. The latest findings from the Symantec and Carbon Black Threat Hunter Team show that Daxin is still operational, after it was found running on a compromised host in Taiwan in 2026. The same machine, belonging to a Taiwan-based subsidiary of a multinational high-tech manufacturer, is also said to have been infected with Stupig ("a.dll" or "kbdus1.dll"). The file name is an attempt to masquerade as "kbdus.dll," a legitimate Microsoft DLL associated with the U.S. English keyboard layout...

Artificial intelligence (AI) is changing offensive security, but it has not changed the standard that matters most: a finding has to be proven before it becomes useful. AI-assisted tools can read code quickly, generate payloads, summarize attack surfaces, explain unfamiliar APIs, and run repetitive testing workflows at impressive speed. That is a real advantage for security teams. It also creates a new kind of pressure, because the industry can now produce more vulnerability-looking output than ever before. The problem is that output is not the same as evidence. A generated report can sound polished, include a severity rating, and even contain a proof-of-concept that looks reasonable at first glance. None of that proves the bug exists in the deployed environment. None of it proves exploitability, impact, or risk. In offensive testing, the hard part has never been writing something that sounds like a vulnerability report. The hard part is demonstrating what is actually true. That d...

Cybersecurity researchers have disclosed details of a previously unreported Internet-of-Things (IoT) botnet framework dubbed TuxBot v3 Evolution that shows signs of being developed with assistance from a large language model (LLM), albeit with not so successful results. "While the AI complied with their request to generate botnet code, it included a safety disclaimer that the developer failed to remove before shipping," Palo Alto Networks Unit 42 said . "Although the LLM clearly aided in constructing the botnet, several functions in the analyzed samples failed to work correctly." The cybersecurity company said a manual code review would have resolved these errors and that it's possible more polished iterations of the malware exist out there in the wild. The botnet framework consists of multiple components: a C-based bot agent that cross-compiles for multiple architectures (e.g., ARM, MIPS, MIPSEL, MIPS64, x86_64, PowerPC, and RISC-V), a Go-based command-...

A single approved marketing tag can quietly load fourth-party code your security team has never seen, granting full access to your forms, customer data, and checkout pages. This on-demand webinar reveals how this Approval Gap forms, and gives your team the blueprint to close it before an auditor, regulator, or attacker finds it first. The Reality of the Approval Gap It's a pattern every security and IT team recognizes: You ran the security review. You approved the vendor. You moved on.

xAI's Grok Build coding CLI was uploading entire Git repositories, full commit history and all, to a Google Cloud Storage bucket run by xAI, not just the files a coding task needed. A researcher publishing as cereblab , testing version 0.2.93 , captured one of those uploads, cloned the git bundle out of the intercepted request, and pulled back a file the agent had been told in plain terms not to open. The upload rode a separate channel from the model itself, and the byte split is hard to argue with. On a 12 GB repo of files the model never read, model-turn traffic to /v1/responses came to about 192 KB while the storage channel to /v1/storage moved 5.10 GiB, a roughly 27,800x gap between what the model needed and what left the machine. That storage upload ran as 73 chunks of about 75 MB, every one returning HTTP 200, and across the researcher's size sweep the volume tracked total repo size. The destination bucket, grok-code-session-traces , is named in the binary and ...

Somewhere right now, a security tool is quietly finding bugs faster than any human can fix them. That's supposed to be the good news. The catch is that the attackers have the same tools, pointed the other way, and they don't file tickets. That's the shape of this week. Trusted code turns on the people who installed it. Old bugs from last year are still landing because the fix sat in a queue too long. Fake installers, poisoned packages, systems left facing the open internet, and helpful little AI assistants running instructions that were never yours. The gap between "patch exists" and "already exploited" keeps shrinking, and nobody's closing it. None of it is exotic. That's what wears you down. Same ordinary mistakes, just happening faster than we can keep up. Here's the full mess, top to bottom. ⚡ Threat of the Week Progress Tells ShareFile Customers to Shut Down Storage Zone Controllers — Progress urged customers to shut down Win...

Give an AI assistant a memory and access to your inbox, and you hand an attacker a way to rewrite what it thinks it knows about you. A single email can trick that agent into saving a false "fact" about the user, hide the change, and quietly steer its answers in later sessions. When it works, the person reads an ordinary-looking reply and never learns their assistant was tampered with. The researchers named the attack  stealth memory injection  and built a tool that writes the emails automatically. The paper, "When Claws Remember but Do Not Tell,"  landed on arXiv on 6 July 2026 . First, what these assistants do A personal agent is an AI assistant that sticks around. Instead of forgetting everything when a chat ends, it keeps notes about you in files: your preferences, your contacts, and what you asked it to do. It reads those notes at the start of every new session, which is why it feels like it knows you. Many of these agents can also act for you, readin...

A new phishing-as-a-service (PhaaS) operation called Forg365 is using a combination of device code phishing , adversary-in-the-middle (AitM) tactics, antibot evasion, artificial intelligence (AI)-assisted lure creation, and post-compromise mailbox operations targeting Microsoft 365 accounts. Distributed via Telegram and costing $400 a month (or $3,800 per year), attack chains leverage phishing lures that make use of legitimate email delivery infrastructure, such as Amazon Simple Email Service (Amazon SES) and Twilio SendGrid, to imitate a redirection chain that blends into regular email traffic before it ends in Forg365-controlled domains. "The panel exposes a mature operator workflow: accounts, links, invitations, OAuth app configuration, redirect links, SVG generation, campaign sending, SMTP profiles, SMTP rotation, AI email generation, token vaulting, account intelligence, keyword alerts, viewer links, and browser-extension support," ZeroBEC said . The email securi...