Today's Cybersecurity Pulse
Anthropic CEO meets White House over federal access to Mythos AI
Anthropic CEO Dario Amodei will meet White House chief of staff Susie Wiles to discuss government access to the company's Mythos AI model, which can discover and exploit zero‑day vulnerabilities. The meeting follows a Pentagon‑imposed blacklist after Amodei refused to lift safety restrictions, while Treasury, intelligence agencies and CISA are already testing the model.
Also developing:
By the numbers: Artemis raises $70M in combined seed and Series A round
From Edge to Enterprise: How the Endpoint Became IT’s Most Strategic Layer and Why Lenovo Is Joining the Conversation at...
Enterprise endpoints have shifted from commodity devices to strategic layers in digital workspaces. As hybrid work, zero‑trust models, and cloud‑first applications proliferate, endpoints now enforce identity, security, and user experience. Lenovo is highlighting this evolution at IGEL’s Now & Next Miami 2026, showcasing AI‑ready laptops, thin clients, and edge workstations paired with IGEL’s read‑only OS. The event gathers CIOs and security leaders to discuss secure, managed endpoint strategies.
New Progress ShareFile Flaws Can Be Chained in Pre-Auth RCE Attacks
Researchers at watchTowr identified two critical flaws—CVE‑2026‑2699 and CVE‑2026‑2701—in Progress ShareFile’s Storage Zones Controller (SZC). The authentication bypass (CVE‑2026‑2699) lets attackers reach the admin interface, while the remote code execution bug (CVE‑2026‑2701) enables malicious ASPX webshell deployment. Chaining the vulnerabilities...
Storm Infostealer Sold as Service, Targets Browsers, Wallets and Accounts
Storm, a new infostealer discovered by Varonis Threat Labs in early 2026, can decrypt Chrome’s App‑Bound Encryption and harvest credentials, session cookies, crypto wallets, and messaging app accounts from Chrome, Edge, Firefox and other browsers. The malware is offered as...
Study Finds 1,748 Exposed API Keys on 10,000 Webpages, Raising SaaS Security Alarm
Security researchers analyzing 10 million webpages identified 1,748 valid API keys on nearly 10,000 sites, many belonging to cloud, payment and developer‑tool services. The leak, largely hidden in client‑side JavaScript, underscores a systemic weakness in SaaS credential management.
Niobium Brings Fully Encrypted AI Workloads to the Cloud with The Fog
Niobium Microsystems is set to launch The Fog, a cloud platform that runs AI and data‑processing workloads on fully homomorphic encryption (FHE) without ever decrypting the data. The service relies on the company’s new mistic Core processor, an FPGA‑based chip...
GitHub Used as Covert Channel in Multi-Stage Malware Campaign
A new multi‑stage malware campaign targeting South Korean users leverages malicious LNK shortcut files that pull PowerShell commands from GitHub repositories. The attackers embed decoding functions directly in the LNK arguments, use decoy PDF documents to mask activity, and establish...
F5 and Forcepoint Partner to Address AI Data Risk and Runtime Security
F5 and Forcepoint announced a partnership that bundles F5’s runtime AI protections with Forcepoint’s Data Security Posture Management (DSPM) for enterprise customers. The joint go‑to‑market approach uses channel partners rather than deep product integration, allowing MSPs, VARs and SIs to...
How North Korean Operatives Get Hired, and How HR Can Stop Them
The U.S. Treasury last month sanctioned six individuals and two entities for operating North Korean IT‑worker fraud networks that generated nearly $800 million in 2024. A joint report by threat‑intelligence firm Flare and IBM X‑Force detailed how these operatives infiltrate American firms...
How North Korean Operatives Get Hired, and How HR Can Stop Them
The U.S. Treasury sanctioned six individuals and two entities linked to North Korean IT worker fraud networks that generated nearly $800 million in 2024. Threat‑intelligence firms Flare and IBM X‑Force detailed how operatives infiltrate U.S. firms using fabricated LinkedIn profiles, tailored...
Boards Are Falling Short on Cybersecurity
Boards increasingly recognize the need for cybersecurity investments, yet their oversight is lagging. A 2024 FBI report shows cybercrime losses jumped 33% year‑over‑year, underscoring the growing threat. The authors identify three core weaknesses: insufficient expertise on boards, superficial risk conversations,...
Verifying AI Agent Intent Becomes Security Priority
Proofpoint is betting big on Intent in the age of AI agents. At RSA Conference 2026, Proofpoint launched Proofpoint AI Security powered by its recent acquisition of Acuvity. The core idea is simple but powerful. Traditional security tools check permissions - Does...
Hackers Shift Focus From Code to Human Exploits
🚨 JUST IN Solana Foundation president Lily Liu states that hackers are now targeting humans instead of code vulnerabilities. 👀 https://t.co/Wjh2kJ7dTT
Former Employee of National Industrial Company Pleads Guilty to Crimes Related to Hacking Computer Networks and Extorting Employees
A former core infrastructure engineer at a New Jersey‑based industrial firm pleaded guilty in federal court to extorting the company and damaging its computer network. In November 2023, the 59‑year‑old Missouri resident initiated unauthorized remote‑desktop sessions, deleted admin accounts, altered passwords,...
Pipelines Pose Greater Security Risks Than Ships
Problem is pipelines are much harder to defend, easier to rupture and cyberattack than ships unless you bury everything far underground which is exceedingly difficult.
Could Native Solana Multi‑Sig Have Stopped the Drift Hack?
If Solana had native multi sig addresses, would the Drift hack even have been possible? Actually curious, not trolling.
Critical Cisco IMC Auth Bypass Gives Attackers Admin Access
Cisco disclosed a critical authentication bypass (CVE‑2026‑20093) in its Integrated Management Controller (IMC) that lets unauthenticated attackers obtain admin privileges on UCS C‑Series and E‑Series servers. The flaw resides in the password‑change function and can be triggered with a crafted...
Anthropic's Code Leak Raises Doubts About Enterprise Security
If Anthropic just leaked their own code (and one of the most valuable pieces of IP on the planet right now)... what makes you think their 'enterprise grade security features' are enough to protect your ideas and data?
Cyber-Physical Security Gaps Demand Attention, Health-ISAC’s 2025 Exercise Series Finds
Health‑ISAC’s 2025 resilience exercise series revealed persistent gaps in coordinating cyber and physical security teams across seven simulated ransomware‑plus‑intrusion tabletop drills. Participants repeatedly reported fragmented communication, delayed escalation, and insufficient joint response procedures. The after‑action report distilled twelve priority observations,...
Google GAs New Ransomeware Protections for Drive, 14x Improvement From Beta
Google announced a new ransomware protection suite for Google Drive that leverages advanced AI to scan and block malicious files before they reach users. The service delivers a 14‑times improvement in detection rates compared with the previous beta program, while...
Why GitHub Developers Are Targeted by Token Giveaway Scams
GitHub developers are increasingly targeted by sophisticated token giveaway scams that masquerade as legitimate project announcements. Attackers exploit developers' public activity, mimicking maintainers, using authentic branding, and leveraging technical language to appear credible. The scams rely on urgency, hidden malicious...
A Quantum Apocalypse Is Coming for the Internet
Google’s quantum research team released a white paper showing it can break 256‑bit elliptic‑curve cryptography using roughly 20 times fewer physical qubits than previously estimated. The breakthrough threatens the cryptographic foundations of most blockchains and many internet security protocols. The article...
US Intelligence Elevates Quantum to AI Threat Level
The U.S. Intelligence Community Just Put Quantum on Equal Footing with AI. And Expanded the Threat Definition https://t.co/BFmWJ9pOmy via @infosec
How ‘Wikipedia of Cyber’ Helps SAP Make Sense of Threat Data
SAP, the world’s largest enterprise‑software provider, is tackling the massive scale of its security telemetry—over 150 TB per month—by partnering with Boston‑based Uptycs. The collaboration introduces Juno AI, an analyst platform that blends artificial intelligence with human expertise to deliver real‑time,...
Possible US Government iPhone Hacking Tool Leaked
Google researchers disclosed a sophisticated iPhone exploit kit called Coruna, which chains 23 iOS vulnerabilities to silently install malware via compromised websites. Evidence points to the toolkit’s origins in the U.S., specifically the Trenchant division of defense contractor L3Harris. Former...
Marquis Fintech Breach Exposes Data of 672,075 Customers in Ransomware Attack
Marquis Marketing Services, a data‑analytics firm used by hundreds of banks, confirmed a ransomware attack that exposed the personal and financial information of 672,075 individuals. The breach, traced to a flaw in firewall provider SonicWall’s cloud backup, has triggered a...
DeFi Platform Drift Freezes Deposits After Hack that May Have Stolen up to $285 Million
Solana‑based DeFi platform Drift announced it is suspending all deposits and withdrawals after confirming an active attack that may have stolen between $136 million and $285 million. The breach sent the DRIFT token down more than 20% and added fresh pressure to...
ISC2 Brings AI Security Into the Heart of Cybersecurity Certifications
The International Information System Security Certification Consortium (ISC2) unveiled new exam guidance that weaves artificial‑intelligence security concepts throughout its entire certification suite. The guidance maps AI topics across more than 50 core exam domains, concluding a three‑year refresh that involved...
Enterprises Are All in on AI for Security but Budgets Aren’t Keeping Pace
Enterprises recognize AI as a cornerstone of modern cybersecurity, with 96% of surveyed security leaders already deploying AI solutions. Yet 85% feel their budgets lag behind the escalating AI‑driven threat landscape, and only 20% claim robust governance frameworks. EY’s survey...
Axios Hack Exposes AI-Coding’s Dependency Problem
Hackers breached the npm account for the widely used JavaScript library Axios, injecting malicious code that was downloaded millions of times before being pulled. The incident follows a similar supply‑chain attack on the LiteLLM PyPI package, highlighting how AI‑coding tools...
SED News: OpenCode, AI Code Vs. Shipped Code, and the LiteLLM Breach
In this episode of SED News, hosts Gregor and Sean discuss the resurgence of ARM CPUs as they move into branding their own chips, driven by the growing demand for local AI agent workloads that favor CPU and memory over...
Rely on Contracts and Legitimate Interest, Not Consent
Most compliance frameworks lean on consent as the primary lawful basis. Consent can be withdrawn at any time. Thailand's PDPA proves it: that's structural vulnerability, not safe harbor. Map your processing to contract performance + legitimate interest first. https://t.co/PTQjD2GryX
FBI Warns of AVrecon Malware Targeting Network Devices Across 163 Countries
The FBI has issued a warning about AVrecon, a modular malware that compromises home and small‑office routers by exploiting unpatched remote‑code‑execution flaws. The malware was used by the now‑dismantled SocksEscort proxy service, which sold access to roughly 369,000 infected devices...
Most CNI Firms Face Up to £5m in Downtime From OT Attacks
A recent e2e‑assure survey of 250 UK critical national infrastructure (CNI) decision‑makers found that 80% of providers could incur OT downtime costs ranging from £100,000 ($132,144) to £5 million ($6.6 million) after a cyber‑attack. Roughly one‑quarter of incidents exceed $1.3 million, and 6%...
Intesa Sanpaolo Missed Unauthorized Access for 2 Years, Regulator Reveals
Italy’s data protection authority disclosed that Intesa Sanpaolo failed to detect an insider’s unauthorized access for over two years. A single employee viewed the records of more than 3,500 customers, leading to a €31.8 million (≈ $34.7 million) fine. Regulators said the bank’s...
Microsoft’s VibeVoice Is Free, Open-Source, and a Compliance Problem Waiting to Happen
Microsoft Research has open‑sourced VibeVoice, a free, locally‑run voice AI that can generate up to 90 minutes of multi‑speaker conversational audio and transcribe 60‑minute recordings with speaker attribution. The system runs on consumer‑grade hardware with roughly 200 ms first‑chunk latency, but was...
Developing A Security Framework For Chiplet-Based Systems
The article outlines a security framework for chiplet‑based systems, emphasizing that each chiplet must possess a verifiable identity tied to a platform‑wide trust chain. It describes two provisioning patterns—certificate‑based external provisioning and silicon‑derived (PUF) self‑generated keys—and explains how both feed...
World First: MACsec IP Receives ISO/PAS 8800 Certification For Automotive And Physical AI Security
Synopsys became the first company to earn ISO/PAS 8800 certification for its MACsec IP, a standard that secures Ethernet communication inside vehicles. The certification, validated by SGS TÜV Saar, confirms that the IP not only protects data integrity but also meets the...
NCSC Warns of Messaging App Targeting Public Sector
The UK National Cyber Security Centre (NCSC) has warned that Russian‑based threat actors are targeting public‑sector personnel through popular messaging apps such as WhatsApp, Messenger and Signal. High‑risk individuals—those with access to or influence over sensitive information—may face login‑code theft,...
Authentication Isn’t Enough: Users Still Fall for Scams
Your customer passed authentication. So why are they sending money to a scammer? https://t.co/Hj99rzCfca https://t.co/wVaFWQnnEq
Massive AI Training Data Leak Threatens U.S. Security
Wow. Incredible amount of SOTA training data now just available to China thanks to @mercor_ai leak. Every major lab. Billions and billions of value and a major national security issue.
Libinput Hit By Worrying Security Issues With Its Lua Plug-In System
Libinput added a Lua‑based plug‑in system in version 1.30 to let developers customize device events. Security researchers have now uncovered two critical flaws—CVE‑2026‑35093, a sandbox‑escape that loads unverified bytecode, and CVE‑2026‑35094, a use‑after‑free bug. Both affect the widely deployed input...
Hashing, Encryption, and Tokenization Explained: How Each One Protects Data Differently
The article breaks down hashing, encryption, and tokenization, explaining how each technique transforms data to protect it. It highlights hashing as a one‑way function ideal for password storage, encryption as a reversible process that secures data in transit, and tokenization...
Federal Cyber Experts Thought Microsoft’s Cloud Was “A Pile Of Shit.” They Approved It Anyway.
In late 2024 FedRAMP granted its cybersecurity seal to Microsoft’s Government Community Cloud High (GCC High) even though internal reviewers called the product “a pile of shit” due to missing security documentation. The agency cited a “buyer beware” notice and the...
Australia Tightens Children’s Online Privacy Rules with New Code
The Australian government, via the Privacy and Other Legislation Amendment Act 2024, has tasked the OAIC with drafting a Children’s Online Privacy Code. The OAIC released the draft this week, opening it for public comment and targeting the 72 million data...
Russia Cracks Down on VPNs, Fines Users and Apple Delists Apps
Russia’s digital minister announced measures to slash VPN usage, including potential fees for heavy international traffic and penalties for non‑compliant platforms. At the same time, Apple has removed several custom VPN and proxy apps from its Russian App Store, prompting...
Instagram Removing End-to-End Encryption: A Precision Harvest
Meta announced it will terminate end-to-end encryption for Instagram direct messages on May 8, arguing the feature sees low adoption. Despite Instagram’s billions of users, the change sparked virtually no public outcry or organized boycott. Critics say the move reflects...
Banks, Telcos to Chuck OTPs, Adopt Silent Authentication
India’s leading private banks and telecom operators are jointly phasing out traditional one‑time passwords in favor of a silent authentication system that validates the mobile number linked to a banking app against the SIM currently active on the device. The...
Anthropic Source‑map Leak Exposes AI Startup Security Gaps
Anthropic confirmed a packaging error that exposed a 512,000‑line source‑map for its Claude Code CLI, revealing internal prompts and feature flags. The leak, quickly mirrored across GitHub, underscores growing security concerns for fast‑growing AI firms and their investors.
Google Study Shows Sub‑1,500 Qubit Quantum Attack Could Break Bitcoin, Slashing Prior Estimates
Google’s Quantum AI team published a whitepaper revealing that breaking Bitcoin’s cryptography may require only 1,200‑1,450 high‑quality qubits, a dramatic reduction from earlier estimates of millions. The finding accelerates concerns over quantum‑resistant cryptography and could reshape the crypto‑security roadmap.
NordLayer Launches Secure Enterprise Browser for SaaS‑Focused SMBs at $8/User
NordLayer introduced a dedicated enterprise browser built for small‑ and medium‑size businesses that run SaaS workloads. Priced from $8 per user per month, the browser adds zero‑trust routing, DLP, Shadow‑IT monitoring and built‑in SSO/MFA, aiming to replace consumer browsers that...