As of Wednesday, August 5, 2026, the AI cybersecurity threat landscape is showing elevated activity across 147 tracked stories. The dominant themes today are AI & LLM Threats (52 stories), Vulnerabilities & CVEs (38 stories), and Data Breaches (18 stories). AI and LLM-based threats are particularly active with 52 stories covering topics such as prompt injection, model exploitation, deepfake campaigns, and AI-assisted attack tooling. 10 CVEs are being tracked across today's stories, including CVE-2026-64531, CVE-2026-59774, CVE-2025-66376, CVE-2018-0171 and 6 more. Each CVE badge below links directly to the NVD advisory. ⚠️ 21 stories involve critical severity, active exploitation, or zero-day conditions. Security teams should prioritize review of flagged items and verify patch status immediately.
Africa’s growing digital economy is exposing governments, businesses and internet users to a rising wave of cybercrime. The continent recorded more than 1.1 billion mobile subscriptions and over $1.1 trillion in digital transactions in 2025, while more than…
“During a routine cyber evaluation, AI agents took sustained, unsanctioned action directed at real people and organisations,” UK’s AI Security Institute (AISI) disclosed on Tuesday. The agents’ actions included an attempted supply-chain attack that saw them…
A memory corruption flaw in the Linux kernel's Open vSwitch datapath gives ordinary local users a path to root on a broad set of default-configured distributions, and a public exploit ships with pre-built records for roughly 800 kernel builds. The…
Kali365 is turning a legitimate Microsoft login into a gateway to corporate data. The phishing kit targets US organizations with attacker-controlled device codes that victims approve on Microsoft's real authentication page. Once access and refresh tokens are…
Hackers stole personal information, medical records, and financial information from the organization’s server. The post 311,000 Impacted by Brown Health Medical Group-MA Data Breach appeared first on SecurityWeek .
An automated system called NOVA read the source code of 3,915 open-source projects over two months and came back with 14,090 vulnerabilities, each one confirmed through the system’s validation pipeline. Vulnerability researchers at Palo Alto Networks’ Unit 42…
The guidelines are the work of the recently launched Open Secure AI Alliance, which now includes 120 organizations. The post Cybersecurity Alliance Drafts SAFE Guidelines for Sharing AI Incident Data appeared first on SecurityWeek .
An unauthenticated attacker can read any file the service account can access on Gitea, the self-hosted Git platform, in versions 1.22.1 through 1.27.0. No login, no repository write access. A public repository and crafted Org-mode markup are enough. The flaw…
Prompt injection remains the most dangerous security threat to LLMs, according to OWASP’s latest Top 10 LLM Applications list
AI Security Institute reports Anthropic and OpenAI models going rogue against real people, organizations, and open source projects. The post AI Agents Targeted Real People and Projects During Cybersecurity Tests appeared first on SecurityWeek .
The flaws can be exploited for remote code execution, authentication bypass, and EncryptInterceptor bypass. The post CISA Warns of Exploited Langflow, N-central, and Tomcat Vulnerabilities appeared first on SecurityWeek .
TP-Link prints the serial number of an Omada router on its packaging and on a label attached to the device. Those numbers run in sequence, and feeding a guessed one to the Omada cloud service returns the matching device’s MAC address and model. Serials…
The malware was designed to steal and exfiltrate secrets, and to propagate itself via stolen NPM and GitHub credentials. The post Over 400 NPM Packages Infected in ChainDrop Supply Chain Attack appeared first on SecurityWeek .
Anthropic and OpenAI models attacked “real people and organizations” during AI Security Institute tests
A phishing campaign impersonating Bank of America (BoA) is underway, trying to trick Windows users into installing ScreenConnect remote access software and then making it difficult to uninstall it. Different traps for Mac and Windows users By claiming the…
Cybercriminals are using a fake Bank of America phishing campaign to trick users into downloading a malicious script that installs ScreenConnect, enabling remote access and persistence on compromised systems
Cloudflare’s Wallets will give AI agents running on its platform a human-readable wallet handle for paying APIs and online content within limits set by their creator. Handle reservations have opened, while the service will become available in the coming…
An agent running Anthropic's Claude Mythos 5 spent 34 hours trying to get a malware dropper merged into a real open-source project during a cyber evaluation by the UK's AI Security Institute. When a bystander publicly warned that the code was malicious, the…
SMOKE#SCREEN uses fake Zoom updates to install ScreenConnect RMM, giving attackers persistent remote access while bypassing defenses. Securonix Threat Research has been tracking an active multi-wave campaign they’ve named SMOKE#SCREEN, in which unknown…
Future AGI is an open-source platform for tracing, evaluating, simulating, and guardrailing LLM agents, licensed Apache 2.0 and self-hostable. Self-hosted instances register with Future AGI on first boot and send an instance ID, a version string, a deployment…
Organizations are building AI systems that combine models, agents and external tools instead of relying on standalone AI, according to Snyk’s latest State of Agentic AI Adoption report. The study analyzed 3,044 enterprise environments and 1.39 million code…
Rogue AI agents from OpenAI and Anthropic have again been caught trying to disrupt servers and software—and leaving instructions for future bad behavior.
A tip from WhatsApp led OpenAI to ban multiple accounts associated with investment scams and human trafficking operations based in Cambodian scam centers.
Swiss Federal IT Agency FOITT says attackers exploited SharePoint flaws to compromise about 200 accounts. Servers are being rebuilt as investigations continue. Switzerland’s Federal Office for Information Technology and Communications, known as BIT or FOITT,…
As Chief of Defence Intelligence, General Sir Jim Hockenhull decided to declassify and publish what London knew of Russia’s plans to invade Ukraine, down to a map of the routes its forces would take.
Build your strategy around answering these questions to ensure employees use AI productively while keeping sensitive data, IP, and agent behavior within the boundaries set for safe AI use. The post Rethinking AI Security: Why CASB and DLP Need an…
INC Ransomware exploits SonicWall SMA 1000 flaws, using calls and emails to pressure victims during extortion campaigns targeting global organizations. Resecurity disclosed that INC Ransomware has emerged as the dominant threat actor exploiting the recently…
A credential-stealing npm worm that first appeared in keyv@6.0.0 spread beyond the Keyv and Cacheable namespaces into hundreds of packages across multiple organizations on August 4, 2026. SafeDep verified 353 poisoned versions across 79 package names in the…
Talos read attacker prompt logs and found guardrails fell to task splitting and ownership claims
This morning, I noticed specific sources "hunting" for vulnerabilities in URLs that I haven&#;x26;#;39;t noticed before. All of these URLs appear to be associated with diagnostic tools: 
The cybersecurity industry has spent decades assuming that offensive capability scales with technical expertise. That assumption is starting to break. Security teams have long estimated risk by ranking attacker sophistication. Nation-state actors sat at one…
Cloud and SaaS are now the preferred operating environments for threat actors, amid a continued shift to identity attacks
Google deleted three AI agent workflows from its Agent Development Kit (ADK) Python repository. Pillar Security showed that a public GitHub issue could manipulate a triage agent into triggering a privileged code-fixing agent. The researchers said the public…
Interpol claims AI is driving a surge in cybercrime in Africa, with related losses doubling
The UK’s Police National Legal Database and Ask the Police service have been breached
Unit 42 uncovered an AI-driven Chinese hacking campaign where DeepSeek autonomously scanned targets, selected exploits, and launched attacks. Researchers at Palo Alto’s Unit 42 got a front-row seat to something they’d only theorized about before: an AI system…
This week kept coming back to permission. A model crossed a boundary. A wallet trusted bad randomness. Webmail kept an intruder around. Public systems, package feeds, hotel networks, and login flows all gave away more than intended. Some of it was clever.…
AI is moving incredibly fast, and every security leader is feeling the pressure to keep up. AI platforms like Claude, Codex and Cursor are already helping security teams write detections, investigate alerts, summarize incidents, and automate repetitive work.…
Three high-severity security flaws have been disclosed in Hugging Face's Diffusers library that could allow crafted model repositories to stealthily execute arbitrary code on machines that load it, opening the artificial intelligence (AI) supply chain to…
Plus: The FBI eyes AI-powered tech to detect future crimes, Russia charges Telegram’s founder, xAI sues to stop a state’s “nudification” ban, and the Democrats learn a lesson about getting scammed.
Both major AI labs’ models broke containment, escaped onto the internet, and hacked other companies. If a human had done that, the law would likely be against them. But a bot?
Most phishing campaigns rely on the fact that the victim is afraid to loose "something": money, access to information, ... Many brands have been impersonated by campaigns but I spotted some phishing emails that focus on AI services like ChatGPT. 
A Chinese-speaking threat actor has been using DeepSeek’s AI models to orchestrate cyber-attacks targeting Asian organizations
Created by legendary hardware hacker Andrew “bunnie” Huang, the badges for this year’s famed security conference aim to push the boundaries of security and transparency.
Anthropic has revealed that Claude AI models compromised third-party organizations
Anthropic on Thursday became the latest artificial intelligence (AI) company to reveal that three of its models, including Claude Opus 4.7, Mythos 5, and an unnamed research model, had breached three unnamed organizations during cybersecurity testing without…
In a review triggered by OpenAI’s Hugging Face incident, Anthropic discovered three of its AI models had breached real-world organizations during third-party evaluations.
The two Chrome updates in June patched more bugs than the 23 updates before them. Now, Google is ramping up its patching schedule thanks to AI-assisted vulnerability discovery.
Security experts have been sounding the alarm for years about the risks of using generic TV boxes that promise unlimited content streaming for a one-time fee, warning that they secretly rent the user's Internet connection out to strangers. But a…
A lot of security still comes down to trusting the wrong screen. This week, that screen might be a login page, an install guide, a recruiter call, or a familiar service behaving slightly wrong. Behind it: reused credentials, exposed systems, quiet loaders,…
Action1 report finds sysadmins overestimated their use of AI in predictions made two years ago
Hidden instructions in a Word document can make Microsoft 365 Copilot rewrite figures in a report, then copy the same instructions into the finished file. Håkon Måløy disclosed the technique on July 28, 144 days after reporting it to Microsoft. In his proof…
Network firewalls are the workhorses of modern cybersecurity. They are trusted to protect the network, blocking malicious traffic and preventing intrusions and breaches. And for decades, network security teams have built controls around a relatively stable…
If the generative AI giant had followed well-known security best practices, it’s likely that its AI agent would never have escaped to the open internet and hacked multiple companies.
Researchers pitted a person against a Claude agent and found that, after a week of texting, the AI chatbot was more effective at creating “exploitable trust” with others.
For now, the use of AI benefits vulnerability research more than vulnerability exploitation, a VulnCheck researcher said
AI is dramatically reducing the barriers to entry for scam phone farm operators, Human Security warns
In a new disclosure, OpenAI says its agent used exposed logins to gain access to at least four “publicly available services” in its unhinged quest to solve a test.
AI-assisted research uncovered Linux kernel use-after-free allowing root escalation
Microsoft has launched a new agentic security system for cyber defenders as well as its first cyber-focused AI model
NVIDIA has launched a new Open Secure AI Alliance to build an “open defense stack for agents”
CREST’s new AI standards are optional add-on requirements for cybersecurity service providers wishing to demonstrate responsible AI usage
Researchers tested top image editing models on Hugging Face and found they could easily create explicit deepfakes—and 1,000 image editing prompts show how people use the software.
The screwup shows how tricky it can be to stop web crawlers from making ostensibly private conversations with AI chatbots entirely too public.
Plus: Russian hackers are trying to steal US nuclear scientists’ emails, the State Department bans known scammers from entering the United States, and more.
OpenAI’s chatbot tool ChatGPT ranked among the top 10 most impersonated brands in phishing attacks for the first time
CoreView research finds that security leadership is concerned about AI Assistant exposing confidential data
Sophos report warns that the rapid adoption of AI by businesses is leaving them vulnerable to a new source of cyber threats
Dolphin X is a new infostealer that uses AI to sort and rank victims, giving cybercriminals a faster way to identify lucrative targets
A new study of organizations which have fallen victim to ransomware suggests the rise of AI-tools being used by hackers is making life harder for defenders
Hugging Face recently disclosed a security breach. OpenAI has now said that it was its AI models which broke containment and hacked Hugging Face themselves
CodeMender actively builds and runs exploits in customer-managed sandboxes to verify if vulnerabilities are truly exploitable
The home appliance giant LG Electronics USA said this week it plans to suspend any apps built for its smart TVs that turn one's television into an always-on residential proxy node. The move comes less than a month after researchers found that more than 42…
<div class="c-page-title__buttons"><a class="c-button" href="https://media.defense.gov/2026/Jul/22/2003965244/-1/-1/1/CSA_RUSSIA_PHISHING_TARGET_ZIMBRA.PDF">Russian State-Supported Cyber Actors Conduct Phishing Campaign Targeting Users of Zimbra Collaboration…
Russian-speaking actor Trim built a commercial offensive AI pentest tool on jailbroken Claude models
FBI warned of deepfake videos of IC3 leadership directing users to spoofed complaint sites
JadePuffer follow-up campaign deployed ENCFORGE locker built to destroy AI model artifacts
A researcher who discovered a critical vulnerability in WordPress has used OpenAI’s latest model to develop an exploit chain
Cybersecurity researchers tested Open AI GPT 5.5’s offensive cyber capabilities – and the results showed how effective a frontier LLM can be for hackers
SANS Institute says governance programs are still nascent even as AI failures and threats grow
The White House announced Gold Eagle to help accelerate the discovery, prioritization and patching of flaws found by AI
Microsoft released fixes for a record 570 CVEs in its July Patch Tuesday update, as experts warn AI is dramatically accelerating vulnerability discovery and increasing patch volumes
Microsoft Corp. today released software updates to plug at least 570 security holes in its Windows operating systems and other software, almost triple the number of vulnerabilities the software giant fixed in its record-smashing Patch Tuesday release last…
The Cybersecurity and Infrastructure Security Agency (CISA) has issued a postmortem on a data leak in which a contractor published dozens of internal CISA credentials -- including AWS Govcloud keys -- in a public GitHub repository for almost six months before…
Researchers at the AI Now Institute developed a proof-of-concept exploit showing common AI tools used for security could backfire
Microsoft has said the volume of Windows security updates is set to grow as it uses AI to find new bugs
Wiz discovered GhostApproval, a symlink flaw in six major AI coding assistants that bypasses approval
Over 70 cybersecurity organizations have signed the CREST AI Charter detailing responsible use of AI for security
Cybersecurity researchers at ESET identify big rise in suspicious and malicious toolsets which put users at risk from cyber-attacks
<p>Russian Government-Sponsored Activity Targets Poorly Configured and Vulnerable Devices Across Critical Sectors</p> <h2><strong>Executive summary</strong></h2> <p>Russian Federal Security Service (FSB) Center 16 cyber actors continue to exploit poorly…
A cybersecurity startup dangling millions of dollars to acquire zero-day security vulnerabilities in popular software is run by a pair of far-right conspiracy theorists and convicted felons whose most recent ventures included fake intelligence companies and a…
Sygnia report details how agentic AI accelerated weeks-long attack to just 72 hours
The National Cyber Security Centre wants to work with AI partners to build a new “Cyber Shield” to defend the UK
Zscaler found sites hiding prompt-injection text to manipulate AI agents into crypto payments
IWF and NCA warn that growing numbers of images and videos are being manipulated into sexual abuse material
Researchers have revealed JadePuffer, the first agentic AI-powered ransomware campaign, highlighting how autonomous agents can automate cyber-attacks
The new classifier in Fable 5 blocks the jailbreak technique that prompted the US export controls “in over 99% of cases”
OpenAI is previewing its GPT-5.6 Sol model to a vetted few at the US government's request
Experts warn the Jaguar Land Rover breach bears hallmarks of Kremlin-backed hackers, citing novel ransomware, strategic timing and efforts to obscure attribution
Cobalt study finds 20-percentage-point drop in number of organizations relying solely on AI automation for testing
LayerX tricked AI browsers including ChatGPT Atlas and Comet into bypassing their guardrails
SentinelLabs found a North Korea-linked macOS backdoor using prompt injection on AI triage tools
New ReliaQuest study reveals the six ways AI is practically being used in attacks today
OpenAI expanded Daybreak with a full GPT-5.5-Cyber release to help defenders patch software flaws
The Five Eyes Alliance has published a rare call to action for organizations facing AI threats
North Korean threat actor Sapphire Sleet has been linked to a supply chain attack targeting Mastra, according to Microsoft security researchers
Working with frontier AI models, this new platform aims to help discovering, prioritizing, validating and remediating code vulnerabilities
A Rust crypto clipper hides behind fake GitHub stars and AI-narrated YouTube videos
Analysis of chatter on underground forums by Sophos finds that hackers fear AI could take work away from them
Nisos infiltrated a North Korean IT-worker fraud cell running on AI interviews and a US laptop farm
The rise of AI-assistants and applications in the enterprise has seen a 93% increase in employees attempting to upload sensitive data, bringing security challenges
Filigran survey at Infosecurity Europe 2026 reveals AI-powered attacks as the top concern, with false positives, alert fatigue and manual processes draining security teams
SANS Institute study finds few SOCs have built AI into defined workflows, despite widespread adoption
Athena is a new an industry coalition to fix the vulnerabilities frontier AI models find before attackers can exploit them
Access to two Anthropic large language models, Mythos 5 and Fable 5, has effectively been banned to any non-US nationals by the Trump administration
Government departments find hundreds of vulnerabilities after testing frontier models
Fake AI guides hide a multi-stage chain that drops AsyncRAT, with signs of AI-assisted coding
Organizations are aware of the challenges that new technologies like AI bring: but cybersecurity staff struggle to make time for the required training during working hours
Tenet Security researchers reveal how new “agentjacking” attacks could trick coding agents into executing arbitrary code
Anthropic unveils Claude Mythos 5 and Fable 5, a restricted-access frontier AI model and guardrailed version for everyone to use
Most dev teams use AI coding assistants but only 30% have full governance in place
OpenAI brings Lockdown Mode and Active Sessions to ChatGPT to curb prompt injection data theft
At Infosecurity Europe 2026, OWASP’s Ariel Fogel warned that prompt injection remains an “unresolved problem” within generative AI architecture
Meta confirms an AI tool vulnerability led to unauthorized access to Instagram accounts after a failure in email verification during password reset
Lloyds Banking Group shared its approach for securing agentic AI workflows, with a mix of hands on experimentation and cross functional governance
The OWASP agentic AI security framework helps organizations assess governance maturity vs adoption and adjust governance as needed
Ox Security field CTO, Boaz Barzel, makes the case for vibe security to tackle AI agent coding risks
Microsoft Detection and Response Team (DART) details how it has uncovered malicious AI applications as cyber criminals manipulate organizations adopting AI tools
A Bugcrowd researcher has unveiled ExploitBench, an independent benchmark of AI models for vulnerability exploitation
Proton uses machine learning models to detect abuse of its services – especially email addresses used by cybercriminals
Inaugural Infosecurity Europe Cyber Startup Award Winner Impresses Panel with Ability Help Prioritize Vulnerabilities in AI era
Trump's executive order invites voluntary pre-release review of frontier AI models
Anthropic widens Project Glasswing access to 150 more firms as patching becomes the bottleneck
The emergence of AI models capable to autonomously find and fix vulnerabilities at scale is having a significant impact on patching management, experts say
Halcyon’s Cynthia Kaiser lifts the lid on the dark web market for AI cybercrime tools
Humans still need to be part of cyber defense, but refusing to deploy AI is no longer optional against AI-enhanced cyber threats, warns Dataminr’s Joe Slowik
Bayer’s security awareness training now focuses on psychological approaches rather than technical methods for detecting social engineering
A threat actor used AI coding tools to build and test EDR evasion malware, Sophos finds
UK organizations are prioritizing AI-driven cybersecurity as 43% cite AI-powered attacks as their top risk, prompting significant investment in advanced threat defense
Top cybersecurity vendors said AI won't replace entry-level – only routine ticket-taking and triage
<div class="SCXW131754345 BCX8"> <div class="OutlineElement Ltr SCXW131754345 BCX8"> <h2><a class="c-button c-button--on-dark"…
<h2><strong>Advisory at a Glance</strong></h2> <table> <tbody> <tr> <th>Title</th> <td>Iranian-Affiliated Cyber Actors Exploit Programmable Logic Controllers Across US Critical Infrastructure</td> </tr> <tr> <th>Original Publication</th> <td><strong>April 7,…
<h2><strong>Summary</strong></h2> <p><strong>Note:</strong> This joint Cybersecurity Advisory is being published as an addition to the Cybersecurity and Infrastructure Security Agency (CISA) May 6, 2025, joint fact sheet <a…
<h2><strong>Executive summary</strong></h2> <p>People’s Republic of China (PRC) state-sponsored cyber threat actors are targeting networks globally, including, but not limited to, telecommunications, government, transportation, lodging, and military…
2.5 million people were affected, in a breach that could spell more trouble down the line.
Researchers uncover a watering hole attack likely carried out by APT TA423, which attempts to plant the ScanBox JavaScript-based reconnaissance tool.
Over 130 companies tangled in sprawling phishing campaign that spoofed a multi-factor authentication system.