Defender's Corner

A year-long study by security awareness platform Pistachio sent 2.47 million simulated phishing emails to 123,000 employees across 1,200 organisations, and found that click rate — the metric most phishing training programmes are judged on — tells only part of the story, according to SecurityWeek.

What actually matters is what happens after the click: 30% of tech and IT workers clicked at least once, and nearly 20% of construction employees handed over credentials after clicking. A click with no credential submission creates no real breach.

What you should do: If your organisation runs phishing simulations, push for reporting that tracks credential submission and suspicious-email reporting rates alongside clicks. If you're an individual, the habit worth building is simple: before entering a password anywhere, check the URL bar first.

An API key (a credential that lets one piece of software authenticate with another) left in a public code repository is a gift to attackers. Bots scan public repositories constantly, and the majority of exposed keys remain active long after they are found.

A security researcher has proposed an open standard called ORKS (Open Revocable Key Standard), detailed at Security Week, that would allow anyone who finds a leaked key to destroy it immediately by submitting it to a revocation endpoint — no account or authentication required.

The standard is not yet widely adopted, but the underlying lesson is actionable right now: if your team uses API keys, set a reminder to rotate them on a schedule and audit whether any have been committed to public repositories using a tool like GitHub Secret Scanning.

The UK's National Cyber Security Centre (NCSC) has published a guide on shadow AI — the use of AI tools that employees adopt without their employer's knowledge or approval, according to the NCSC blog.

One survey cited in the piece found that 71% of employees use AI tools their employer has not approved. The risk is straightforward: when you paste work data into a consumer AI service, that data may be stored, retained, or used to train the service — outside any security controls your organisation has in place.

The NCSC's advice is practical: before using any AI tool for a work task, ask whether it is on your employer's approved list. If you are not sure, ask. If no approved option exists for your needs, raise it with your IT or security team rather than improvising. The goal is not to stop using AI — it is to use it without handing your company's sensitive data to an unknown third party.

HaveIBeenPwned is a free service that tells you whether your email address or phone number has appeared in a known data breach. It is run by respected security researcher Troy Hunt and is widely trusted.

With 8.8 million records from the Manchester Airports Group breach now in its database, this is a good moment to check. Go to haveibeenpwned.com, enter your email address, and see which breaches it appears in. You can also set up free alerts so you are notified automatically if your details appear in future leaks.

If your email is listed, update passwords on affected accounts and switch on two-factor authentication (a second verification step beyond your password) wherever it is offered.

Anthropic has warned Claude users that infostealer malware — software that quietly copies saved passwords, browser cookies, and login credentials — was used to hijack accounts and drain usage limits, according to SecurityWeek. The malware arrives via unofficial downloads or fake apps, not through Claude itself.

If your Claude usage limits appeared to drain without you doing anything, that is a warning sign. Anthropic has signed out affected sessions and removed saved payment methods as a precaution.

The practical step: run a full scan with Malwarebytes — a free, widely trusted tool — before re-adding any payment details to Claude or any other service. Then change your passwords for any accounts that were logged in on the affected device. This applies whether or not you use Claude: infostealers target every saved credential on your machine.

The UK's National Cyber Security Centre has issued a warning about increased attacks on operational technology (OT) systems — the industrial software and hardware that runs power grids, water treatment, and manufacturing, according to the NCSC advisory.

Their first recommendation applies to everyone, not just industrial operators: know what you have exposed to the internet.

For home and small business users, the practical version is this: log into your router's admin panel and check which devices are reachable from outside your network. Most routers have a "port forwarding" or "exposed services" section. If you see something you did not intentionally open up — a printer, a camera, a server — close it.

If you manage any internet-connected device for work, change its default password today. Default credentials are the first thing attackers try, and they work more often than anyone should be comfortable admitting.

WhatsApp has rolled out several security improvements worth enabling now, according to Security Week.

Two-step verification (2SV — a second layer of protection beyond your SIM card) has been upgraded from a six-digit PIN to a full password that can include letters, numbers, and special characters. Longer and more complex is better. Go to Settings → Account → Two-step verification and update yours.

Passkeys (a login method that replaces passwords with a cryptographic key stored on your device) can now be added in multiples, which is useful if you switch between iOS and Android. Android users also get a new caller context feature that shows the country and any shared group membership for unknown callers — a quiet but useful tool against number-spoofing scammers.

The Uber fine (covered below) is a useful reminder that EU law gives individuals the right to challenge automated decisions — and similar rights exist in the UK, Canada and California under their respective data protection laws.

If a company tells you that an automated system has denied you a service, suspended an account, or made a significant decision about you, you can ask for a human review. In the EU and UK, this right is explicit under GDPR and UK GDPR. To exercise it, contact the company in writing, reference "automated decision-making," and request that a person reviews the outcome.

It will not always work quickly. But companies that ignore such requests face exactly the kind of regulatory attention Uber just received. Know the right exists — it is worth using.

Anthropic — the company behind the Claude family of AI models — has expanded a programme called Project Glasswing, which gives security teams access to its most capable AI model, Mythos 5, according to Security Week. It has also launched a $35 million fund to help open source projects find and fix vulnerabilities.

For most readers, the practical takeaway is this: Claude Security, available now in public beta for Claude Enterprise users, can scan your codebase, flag vulnerabilities with severity ratings, and suggest fixes — with a human required to approve anything before it deploys.

If your organisation uses Claude Enterprise, check whether Claude Security is enabled for your team. If you maintain open source software, the Defender Advantage Fund (0xDAF) is worth looking into for grant support.

Email filters were built to catch known threats. AI-generated phishing emails are designed to look like normal business correspondence, personalised using public information from LinkedIn and company websites. A Harvard Business Review study found AI-crafted spear phishing (targeted, personalised phishing) hit a 54% click-through rate. Filters alone will not cut it.

The article from Bleeping Computer makes a useful point: assume some phishing will always reach the inbox. The better question is what you have in place once it does.

What you should do: Pair your email filter with endpoint detection software and identity monitoring. Enable multi-factor authentication on every account so that a stolen password alone cannot open the door. When in doubt about an email, call the sender directly using a number you already have.

Microsoft has removed WMIC (Windows Management Instrumentation Command-line) from Windows 11 versions 24H2 and 25H2, according to Bleeping Computer. WMIC is a built-in Windows utility that attackers have long repurposed to delete backups, disable antivirus software, and dig deeper into compromised systems — all without triggering alerts, because the tool was signed by Microsoft itself.

Its removal means a whole category of well-established attack techniques simply stops working on updated systems.

What to do: Open Windows Update and make sure you are running the latest version of Windows 11. If you manage Windows machines for a small business, check that automatic updates are enabled across all devices. This one costs nothing and closes real doors.

GitHub's Dependabot is a built-in tool that automatically scans your code's dependencies (the third-party packages your project relies on) and alerts you when something dangerous turns up.

Until this month, Dependabot's malware alerts only covered npm, one of the most popular package registries. That meant poisoned packages from PyPI, Maven, RubyGems, NuGet, Go, crates.io, and PHP Composer went undetected. As reported by Help Net Security, all eight ecosystems are now covered.

If you manage any GitHub repository, enable Dependabot malware alerts now. Go to your repository, click Settings → Security & analysis → Dependabot alerts, and switch it on. It is free, takes thirty seconds, and runs across more than 30 million repositories already.

The macOS Screen Sharing vulnerability being actively exploited right now is a good reminder that built-in remote access features are a liability when left on by default. Screen Sharing, Apple's remote desktop tool, only needs to be reachable from the internet to become a target — and many users have no idea it's enabled.

Disabling it takes thirty seconds. On your Mac, open System Settings → General → Sharing, find Screen Sharing, and switch it off. If you do need remote access to your Mac, consider using it only over a VPN rather than exposing port 5900 directly to the internet.

More broadly, go through everything listed under Sharing in System Settings. Remote Login, Remote Management, and AirPlay Receiver are all features worth disabling if you don't actively use them.

WhatsApp is rolling out a new feature called Scam Alert, according to Bleeping Computer. It runs a machine learning model directly on your device to flag messages from unknown contacts that match known scam patterns.

Nothing leaves your phone for analysis. No message content is sent to WhatsApp, Meta, or anyone else. If the model spots something suspicious, you will see a prompt to block, report, or continue. If it gets it wrong, you can mark the chat as trusted.

The feature is in limited beta right now but is expected to roll out more broadly. To check if it is available: open WhatsApp, go to Settings → Privacy and look for Scam Alert. Enable it if you see it.

Scam messages via messaging apps are one of the most common ways people are defrauded. This is a free, privacy-respecting layer of protection that takes thirty seconds to enable.

Signal has launched Automatic Key Verification, a new feature described by Bleeping Computer that protects against man-in-the-middle attacks (where someone intercepts your messages by quietly swapping out your contact's encryption key for their own). Previously, confirming a contact's key required meeting in person or using a separate channel. Now Signal uses Cloudflare and Trail of Bits as independent auditors to verify that the key associated with a phone number has not been tampered with.

To enable it: open Signal, go to Settings > Privacy > Advanced, and toggle on Automatic Key Verification. A green "Encryption verified" badge will appear when a chat is confirmed clean.

This matters because Russian state-sponsored hackers previously compromised Signal accounts by abusing the Linked Devices feature. This feature closes a gap they exploited.

Google announced that Chrome's anti-abuse systems blocked over 7 billion unwanted Android notifications per day in Q1 2026, according to Bleeping Computer.

Notification abuse is a real attack channel: scammers use browser notifications to deliver phishing links and fraudulent payment requests. Chrome now automatically revokes notification permissions from inactive or suspicious sites and lets you review what it has removed.

What to do right now: On Android, open Chrome and go to Settings > Notifications. On desktop, go to Settings > Privacy and security > Site Settings > Notifications. Review which sites have permission to send you notifications and revoke anything you do not recognise. Chrome's Safety Hub will also show you permissions it has already revoked automatically — check that list too. If a site you removed turns out to be legitimate, you can restore it.

This takes about two minutes and cuts a surprisingly effective scam delivery channel off at the source.

If you or your team write Python code, your encryption just got a significant upgrade available with a single install. The pyca/cryptography library now includes support for ML-KEM and ML-DSA, the two post-quantum cryptography standards finalised by NIST (the US National Institute of Standards and Technology). Post-quantum cryptography refers to encryption algorithms designed to resist attacks from quantum computers, which are expected to eventually break most encryption in use today.

You do not need a quantum computer bearing down on you to act. The point of switching now, as the project puts it, is precisely that there is no emergency yet — which means time to migrate carefully rather than frantically.

To get it: pip install cryptography (version 44 or later).

If you are not a developer, ask whoever manages your software stack whether your tools are on the roadmap for post-quantum upgrades.

GitHub's Dependabot is a built-in tool that automatically scans your project's dependencies — the third-party packages your code relies on — and alerts you when something is outdated or dangerous. Until now, its malware detection only covered npm (a package registry for JavaScript developers). That changed this month, according to Help Net Security.

Dependabot now draws from OpenSSF's malicious-packages database, a public feed covering eight ecosystems including PyPI (Python), Maven (Java), RubyGems, NuGet (.NET), Go, crates.io (Rust), and PHP Composer.

If you use GitHub for any development work, enabling Dependabot malware alerts takes one minute and adds an automatic watchdog across your entire dependency chain.

What you should do: Go to your GitHub repository or organisation settings, find the Security section, and enable Dependabot malware alerts. It is free and takes effect immediately.

The RovoBlast finding, presented at DEF CON 34, illustrates a risk that applies well beyond Atlassian: AI assistants with broad integration access can be weaponised to exfiltrate data automatically once an attacker controls their input.

Varonis researchers recommend a concrete set of steps that any team using an enterprise AI tool can apply right now, according to SecurityWeek:

1. Audit which systems your AI assistant can access and remove any integrations not actively used.
2. Block AI access to sensitive departments — HR, legal, and finance — unless there is a specific, reviewed business need.
3. Disable autonomous browsing or multi-step agent features if your team does not rely on them daily.
4. Enable activity logging for your AI assistant and review it regularly for unusual data access patterns.

Think of it like giving a new intern a master keycard on day one — useful in theory, but you would probably hand them the specific keys they actually need instead.

The Swiss government breach is a useful reminder: it is not just about patching. It is about knowing what is exposed in the first place.

Microsoft SharePoint — the document storage and collaboration platform used by thousands of organisations — often ends up with external internet access enabled by default, even when that access is not needed. That is unnecessary attack surface.

Do this today:

1. Open your Microsoft 365 admin centre and check SharePoint's external sharing settings under Policies → Sharing
2. If external sharing is not required for your team, set it to "Only people in your organisation"
3. Confirm your July 2026 SharePoint patches are applied via Microsoft 365 admin → Health → Message Centre

Takes under 10 minutes and meaningfully reduces your exposure.

SharePoint admin sharing settings — Microsoft Docs

Microsoft has updated its free Zero Trust Assessment tool to include a dedicated AI security pillar, according to Help Net Security.

Zero trust (a security model built on the principle of "never trust, always verify" — no user or system is trusted by default, even inside your own network) is increasingly relevant as organisations deploy AI agents with access to sensitive tools and data.

The tool automatically scans your Microsoft security configuration, identifies gaps, and now produces a prioritised roadmap covering identity, devices, data, network, and AI systems. It is free for Microsoft environment users.

If your organisation uses Microsoft 365 or Azure, run the assessment at Microsoft Security. Pay particular attention to the AI pillar if you have deployed any Copilot or AI agent tools. The results will tell you exactly what to fix first.

Cloudflare has announced AI agent wallets with built-in spending controls, according to Help Net Security. The feature lets you create a Virtual Wallet for any AI agent you run, with hard caps on how much it can spend, which services it can pay for, and per-transaction limits — all controlled by you through an Account Wallet.

This directly addresses a real risk: AI agents given open access to APIs and payment methods can rack up unexpected charges or be manipulated into spending beyond their intended scope.

Handle reservations are open now at Cloudflare. If you use AI agents for any work task, reserve a handle and review what spending permissions your current agents already have. Giving an AI agent uncapped access to APIs or services is roughly equivalent to handing a new employee an unlimited corporate card without a receipts policy.

The full announcement is at helpnetsecurity.com.

NuGet is Microsoft's software package repository for .NET developers — think of it as an app store for code libraries. Starting 17 August 2026, new API keys (passwords that allow developers to publish software packages) will expire after 30 days instead of 365, according to Help Net Security.

Why does this matter? A stolen long-lived key is a skeleton key that stays valid for a year. A recent attack used stolen credentials to push a malicious package that was downloaded 6,000 times in 36 minutes. Shorter lifetimes cap the damage window.

If you are a developer publishing to NuGet.org, audit every workflow that uses an API key and replace any key created before 17 August before 1 November 2026. Better still, migrate to NuGet Trusted Publishing — it uses short-lived, automatically expiring credentials instead of permanent keys.

Malware on Windows and macOS can quietly write fake enterprise policy keys to your system, then use those keys to force-install browser extensions that replace your search engine or new tab page — and Chrome treats them as administrator-approved, making them impossible to remove. Google is preparing a fix, as reported by Bleeping Computer.

The feature is not live in stable Chrome yet, but when it ships, it will automatically block these fake policy installs on personal devices and uninstall hijacking extensions that were previously locked in.

What to do right now: If your Chrome shows "Managed by your organisation" and you do not work on a company device, that is a red flag. Type chrome://management in your address bar. If it says your browser is managed but you do not recognise the organisation, run a malware scan with Malwarebytes Free. Keep Chrome updated so the new protection arrives as soon as Google ships it.

A live AMOS (Atomic macOS) stealer campaign is targeting Mac users through fake "macOS toolkit" websites. The trick is simple: a site presents copy-paste instructions that look like a legitimate software setup step. You open Terminal (Mac's command-line tool), paste the text, and unknowingly install a credential-stealing app that harvests passwords, browser data, and cryptocurrency wallet files.

According to SANS ISC, the malware phones home to a remote server and begins collecting data immediately.

The fix is equally simple: never paste anything into Terminal from a website unless you have read it character by character and understand what it does. Legitimate software installers do not require this. If a site tells you to open Terminal and paste something, close the tab.

If you have already done this, change your passwords and check your accounts for unusual activity today.

CISA issued an urgent alert after attackers disrupted more than 30 community water systems in Minnesota by targeting internet-exposed PLCs (programmable logic controllers — the industrial computers that manage physical equipment like pumps and valves), according to Bleeping Computer.

Attackers changed passwords to lock operators out, modified IP addresses to disconnect devices, and forced some utilities to switch to manual operations.

CISA's primary instruction is simple: remove these devices from direct public internet exposure immediately. If that is not possible, place them behind a VPN (a secure, encrypted access tunnel) or gateway device. Change all default passwords and restrict access to a known list of approved IP addresses.

This matters beyond utilities. The same principle applies to any remotely accessible device you manage — routers, smart home hubs, or office equipment. Anything reachable from the public internet without a barrier is a target.

What you can do today: Check whether any devices in your home or workplace are directly accessible from the internet. Your router's admin interface is a good place to start.

Most security advice focuses on keeping attackers out. A real incident documented by Huntress, covered on Bleeping Computer, shows why what happens next matters just as much.

In the case Huntress investigated, the attacker gained entry through a SQL injection flaw (a vulnerability where a website passes user-typed input directly into a database without checking whether it is safe), then quietly created a new admin account, enabled remote desktop access, disabled Windows Defender, and installed malware on the web server — all before anyone noticed.

The takeaway for non-technical readers: attackers rarely rush. They settle in.

What to do: Check your active user accounts periodically and remove any you do not recognise. If you manage a website, ask your hosting provider whether input validation and web application firewall protections are in place. Huntress offers free resources for small businesses at huntress.com.

Research from Vanson Bourne surveying 600 senior IT security professionals found that 73% of organisations admit they would not be fully prepared if a serious cyberattack happened tomorrow, according to The Hacker News. The biggest gap is not tools — it is coordination. Legal, communications, and executive teams often arrive late when speed matters most.

The fix is a tabletop exercise: a scripted, hypothetical attack scenario that your team walks through together without any real systems at risk. Think of it as a fire drill for a breach.

You do not need a security consultant to start. CISA publishes free tabletop exercise packages at cisa.gov/resources-tools/services/cisa-tabletop-exercise-packages. Pick a scenario that fits your organisation, schedule two hours, and get legal and leadership in the room. The goal is to find the gaps before an attacker does.

CISA, the FBI, and their Australian counterparts have published free guidance called CI Fortify — Advice for Isolating Vital Systems, aimed at helping organisations disconnect critical systems from their networks during an attack while keeping essential services running. Read it here via Bleeping Computer.

The core exercise it recommends translates directly to smaller operations and even households: map every connection your most important systems have to the outside world, and decide in advance exactly where you would cut them if something went wrong.

For a small business, that means knowing which systems matter most, which vendors and cloud services they depend on, and how long you could operate manually if those connections were severed. For individuals, it is worth knowing where your router admin panel lives, what your ISP support number is, and whether your most critical files exist somewhere offline.

Knowing the answer before an incident is the difference between a plan and a panic.

A supply chain attack (where hackers poison a software package that developers trust and use, sending malicious code downstream to every project that installs it) is one of the most effective ways to compromise thousands of organisations at once.

GitHub and PyPI — the Python Package Index, the main repository where Python software packages are published — have both introduced new protections, according to Security Week.

GitHub's automated dependency tool, Dependabot, now waits three days before updating your project to a brand-new release. That gap gives security researchers time to catch a poisoned version before it reaches your codebase. PyPI is now blocking new files from being added to releases older than 14 days, closing a gap where an attacker with a stolen publishing token could quietly poison a long-trusted package version.

What to do: If your team uses Dependabot, check that your dependabot.yml configuration is not overriding the new cooldown behaviour. If you maintain Python packages on PyPI, no action is needed — the protection applies automatically.

CTM360 researchers documented an evolved insurance phishing campaign, reported by The Hacker News, where attackers no longer just collect your credentials to use later. Instead, a fake insurance portal acts as a live middleman: as you type your username and password, the attacker simultaneously submits them to the real insurance site in real time, completing the login before you realise anything is wrong.

The primary entry point was fake Google Ads promoting insurance quotes and renewals.

CTM360's report does not identify a single technical control that defeats this entirely, but the most reliable protection it implies is avoiding the attack's starting point. Before entering any credentials on an insurance or financial portal, type the organisation's URL directly into your browser rather than clicking any advertisement or search result link. Check the full URL in your address bar carefully before submitting anything.

Meta has launched Facebook Verified, a free identity badge for personal accounts, according to Help Net Security. To get it, you record a short video selfie. Facebook checks it against your existing profile photos to confirm you're a real person.

The badge appears on your profile and in Marketplace, Dating, and Groups — exactly the places where AI-generated fake accounts are most dangerous. It costs nothing and takes a few minutes.

If you use Facebook Marketplace or Dating, this is worth doing now. A verified badge signals to other users that you passed an identity check — and it makes your account a less attractive impersonation target.

To get started: go to your Facebook profile settings and look for the Verified badge option. You must be 18 or older with an account in good standing.

Microsoft is changing how Windows enterprise activation works, replacing a software-only trust model with hardware-backed verification using TPM (Trusted Platform Module — a dedicated security chip built into most modern computers that stores cryptographic keys and confirms the machine hasn't been tampered with), according to Help Net Security.

The change affects Windows Key Management Service (KMS), the on-premises tool organisations use to activate Windows across large fleets of machines. From August 2026, Windows Server 2025 will flag whether your KMS host is ready for hardware-based security. Mandatory enforcement arrives with the next Windows Server LTSC release.

What to do now: If you manage Windows servers, identify your KMS hosts and confirm whether the physical machines have TPM support. If they don't, hardware upgrades may be required before enforcement kicks in. Loop in your IT team and review Microsoft's rollout schedule ahead of the August readiness messaging.

PyPI (the Python Package Index) is the central repository where developers download Python software libraries. As of this month, it now rejects any new files uploaded to a release older than 14 days.

This matters for non-developers too. If an attacker compromises a developer's credentials, one classic move is to slip malicious code into an older, trusted package version. Users downloading what they believe is a stable, long-tested release would get the poisoned version instead.

The new rule shuts that window. Older releases are now frozen. Attackers cannot retroactively tamper with them even if they steal the right credentials.

You do not need to do anything to benefit from this change. It is already live for all packages on PyPI.

What to do: If your organisation's software teams use Python packages, ask them to confirm they are pulling dependencies from PyPI directly rather than caching old versions locally from before July 2026.

ServiceNow is enterprise software used by organisations to manage IT helpdesk tickets, HR requests, and internal knowledge bases. Security firm Bishop Fox found that 31% of the 166 ServiceNow instances they tested during authorised engagements returned records or confirmed record counts to completely unauthenticated requests, according to Help Net Security.

Bishop Fox released Snowpick, a free open-source scanning tool that checks your own ServiceNow instance for this exact misconfiguration. It works by grabbing the public session token your portal issues to anyone who visits the login page, then using it to probe default widgets and API endpoints for accessible data.

If you manage a ServiceNow instance, run Snowpick against it before someone else does. The tool is available on Bishop Fox's GitHub. If you are an employee, ask your IT team whether your organisation's ServiceNow portal has been audited for public data exposure.

Amazon GuardDuty — a managed threat detection service that monitors AWS cloud accounts for suspicious activity — has launched an AI-powered investigation agent, now in public preview and free to use during the preview period, according to Help Net Security.

When a security alert fires, the investigation agent automatically analyses the finding, maps it to the MITRE ATT&CK framework (a standardised catalogue of attacker techniques), assesses risk and confidence levels, and produces actionable remediation steps. Think of it as the first analyst on the scene writing up the incident report before your team even opens their laptop.

It is available in 10 AWS regions, limited to 10 investigations per account per day during preview.

What you should do: If your organisation runs workloads on AWS and has GuardDuty enabled, turn on the investigation agent in your GuardDuty console today. It costs nothing during preview and reduces the window between alert and action.

Microsoft has released Dusseldorf, a free open-source platform for out-of-band application security testing (OAST) — a method of catching vulnerabilities that don't leave obvious traces inside an application but cause the app to quietly make unexpected outbound network connections.

It detects flaws like SSRF (server-side request forgery, where an attacker tricks a server into making requests on their behalf), cross-site scripting, and XML external entity injection — all by watching what the application reaches out to at the network edge.

Dusseldorf runs DNS, HTTP, and HTTPS listeners on a domain you control. Every unexpected request gets logged. It is designed for private deployment and comes with a graphical interface and API.

It is free on GitHub and built for both local lab setups and Azure cloud environments.

Get it: Microsoft Dusseldorf on Help Net Security

The 2026 SANS AI Survey found that 78% of security practitioners now use generative AI in their daily work — but the same research, covered by Help Net Security, highlights that human judgement remains the most effective check against AI-driven attacks.

Separately, researchers at Tracebit found that planting "context bombs" (deliberately misleading information seeded into environments to confuse AI agents) significantly slowed automated attackers from completing their objectives.

You do not need a security team to apply the core principle: slow down and question anything that creates urgency. Phishing and social engineering attacks, including the ClickFix technique spreading ACR Stealer, work because they pressure you to act before you think.

What to do: Before clicking, pasting, or downloading anything unexpected, pause for ten seconds and ask: did I request this? If the answer is no, close the window.

Apple has published a warning that scammers are using FaceTime to impersonate banks and Apple itself, per Help Net Security.

The playbook goes like this: a caller spoofs (fakes) a trusted number, drops some personal details early to sound legitimate, then creates urgency so you stay on the line. The goal is to get you to share your password, a two-factor authentication (2FA) code, or disable security features entirely. Some callers instruct victims to turn off 2FA or Stolen Device Protection, claiming it will stop an attack. It will not. It opens the door to one.

Apple is clear: it will never ask you to share a password, a 2FA code, or to accept a 2FA prompt over the phone.

What to do: Hang up. Call the organisation back on a number you look up yourself. If you receive a suspicious FaceTime call from someone claiming to be Apple or a bank, take a screenshot and send it to reportfacetimefraud@apple.com.

1Password has released a beta integration with Claude that lets the AI assistant complete browser tasks requiring a login — without ever seeing your actual password, according to Help Net Security.

Here is how it works: when Claude hits a sign-in page during a task, it requests the credential from 1Password. You approve the request using Touch ID or another biometric method. The password is filled in through a secure channel that never enters Claude's memory or Anthropic's systems. Once the task ends, the permission resets.

This is available to paid Claude subscribers (Pro, Max, Team, or Enterprise) on macOS with the 1Password desktop app and browser extension installed.

If you use Claude for browser-based tasks, this is worth enabling. It keeps your credentials out of AI context windows entirely — which is exactly where they should not be.

Single sign-on (SSO) is the system that lets you log in once and stay logged into multiple apps and services without re-entering your credentials. Microsoft has quietly made this easier to manage for IT teams, according to Help Net Security.

Windows 11 now supports a registry-based policy that lets administrators automatically accept SSO permission prompts on managed devices linked to Microsoft Entra ID (Microsoft's cloud identity platform). This removes friction for employees while keeping IT in control of what access is granted.

If you manage devices at work, you can deploy this through Group Policy, Microsoft Intune, or any compatible mobile device management tool.

What you should do: If you are an IT administrator running Windows 11 version 24H2 or 25H2 with Entra ID, review Microsoft's SSO policy documentation and test the registry setting in a small device group before rolling it out broadly. Less friction for users, more consistency in access control for you.

Microsoft is making passkeys the default login method for Entra ID (an enterprise identity service used by millions of organisations to manage employee logins) starting September 2026, according to Bleeping Computer. SMS and voice authentication codes will be fully retired in February 2027.

Passkeys work like a secure key stored on your device — no code to intercept, no phishing link that works. Microsoft reports that AI-assisted phishing campaigns now achieve click-through rates of 54%, compared to roughly 12% for traditional attacks. SMS codes are vulnerable to exactly that.

If you use Microsoft work accounts: ask your IT team whether you are on SMS-based login. If you are, register a passkey or Windows Hello before the September rollout. Microsoft's step-by-step setup guide is here.

Starting September 1, 2026, Microsoft will begin making passkeys the default sign-in method for Microsoft Entra ID (the identity system used by Microsoft 365 and most corporate Microsoft accounts), according to Help Net Security.

A passkey is a login credential stored on your device — your phone or laptop — that uses biometrics or a PIN instead of a password. It cannot be phished, because there is nothing to hand over to a fake website.

If your organisation uses SMS codes for multi-factor authentication (MFA), you will be prompted to set up a passkey the next time you log in. By February 2027, SMS-based MFA will no longer be accepted without registering a passkey first.

You do not need to do anything right now — Microsoft will prompt you. But when that prompt appears, accept it. Passkeys are meaningfully more secure than SMS codes, which can be intercepted.

A SentinelOne investigation into a multi-year espionage campaign against Pakistani law enforcement reveals a technique worth understanding: attackers disguised malware as a legitimate software update for a portal already trusted by staff, according to The Hacker News.

The implant was named to look like a routine plugin for a complaint management system. Staff had no reason to question it.

This is called a trojanised update (legitimate-looking software that secretly contains malicious code), and it works precisely because users trust update prompts from systems they use daily.

What you can do: Treat unexpected software updates with the same scepticism you would give an unexpected email attachment. If a portal or internal tool prompts you for an update you were not expecting, verify it with your IT team before clicking. Legitimate updates from your organisation will always be communicable through an official channel. When in doubt, ask first.

A study of 281 free Android VPN apps found widespread problems including traffic leaks, unencrypted data transmission, and hidden tracking, according to Bleeping Computer. A VPN is meant to create a secure, private tunnel for your internet traffic — but an untrustworthy one can expose exactly what you were trying to protect.

Free VPNs with no clear business model often fund themselves through your data.

What to do: If you use a VPN on Android, stick to well-reviewed paid services or widely trusted free options with published transparency reports. Check your current app's permissions — a VPN has no business accessing your contacts or microphone. If you are not sure whether your VPN is legitimate, removing it is safer than keeping it.

GitHub released npm 12 this week, and one change directly addresses the kind of supply chain attack described in today's breach story. According to The Hacker News, install scripts — small programs that run automatically when you download a code package — are now disabled by default.

Previously, a malicious package could execute code the moment a developer ran an install command. Now, those scripts must be explicitly approved first.

If you are a developer: upgrade to npm 12 now. Run npm approve-scripts --allow-scripts-pending to review any scripts in your current project before approving them. Think of it like previewing a contract before signing it, rather than having it auto-signed the moment it arrives.

Not a developer? Forward this to whoever manages your team's software builds. This one matters.

Get npm 12 at npmjs.com.

The Friendly Fire research above makes one defensive tool directly relevant today: AWS has released the Claude Apps Gateway, described by Help Net Security. It is a self-hosted control plane for organisations using Claude Code (Anthropic's autonomous coding tool) in a team setting.

Instead of each developer holding their own long-lived credentials, the gateway issues short-lived tokens that expire within an hour. It centralises policy controls and enforces tool permissions across your whole team. Organisations managing multiple developers can use it to limit which tools agents can access and revoke access instantly by removing a developer from your identity provider.

This directly limits the blast radius if an agent is manipulated into executing malicious code. Fewer permissions mean less damage. Find it via Help Net Security.

Langflow is a drag-and-drop tool for building AI agent pipelines — popular with developers connecting large language models to external services. It has been having a rough year.

CISA (the US Cybersecurity and Infrastructure Security Agency) has ordered federal agencies to patch CVE-2026-55255, an actively exploited flaw in Langflow, according to Bleeping Computer. Attackers who exploit it can access other users' data flows and harvest cloud credentials and compute resources for botnet use.

If you or your team runs a Langflow instance, update it immediately. Attackers are already exploiting this, and the prize — live LLM API keys and cloud credentials — is exactly the kind of access that funds further attacks.

Check your Langflow version at langflow.org and apply the latest update. If your instance is internet-facing, take it offline until patched.

Microsoft is enabling its Windows settings backup tool by default for organisations running Windows 11 26H2, according to Bleeping Computer. The feature, available on Microsoft Entra-joined devices, automatically backs up a user's Windows settings so they can be restored after a device is reset, replaced, or wiped.

This matters from a security standpoint: when employees lose a device or face a ransomware recovery, having settings backed up cuts recovery time significantly and reduces the window of exposure.

What you should do: If your organisation uses Microsoft Entra (formerly Azure Active Directory), check whether the backup policy is configured intentionally. IT administrators can manage the setting through Microsoft Intune or Group Policy. If your organisation operates in an EU Digital Markets Act region, the default-on behaviour does not apply — but enabling it manually is still worth considering.

Most people send hundreds of emails a week without realising the protocol was designed with almost no authentication at all. That gap is one of the most commonly exploited entry points for phishing and brand impersonation attacks, according to a Help Net Security feature on email security strategy.

The practical fix for businesses and individuals running their own domains is DMARC (Domain-based Message Authentication, Reporting and Conformance — a standard that tells receiving mail servers what to do when an email claiming to be from your domain fails authentication checks). Without it, anyone can send email pretending to be you.

Check whether your domain has DMARC configured using MXToolbox's free DMARC lookup — enter your domain and it will tell you in seconds. If it shows no record, ask your email provider or IT contact to set one up. Start with a policy of p=none to monitor, then move to p=reject to block fakes entirely.

The PolinRider campaign described above spreads through a VS Code feature called task files. These are configuration files that can run code automatically when you open a project folder. Most developers never review them, which makes them a perfect hiding spot.

Here is what to do right now. In any project you have cloned or installed recently, open the .vscode folder and look for a file called tasks.json. Check whether any task includes "runOn": "folderOpen". If it does and you did not put it there, treat it as suspicious.

You can also disable automatic task running entirely. In VS Code, go to Settings and search for "task auto detect" to review what runs without your explicit permission.

This takes five minutes and requires no extra software.

A joint operation involving Google, the FBI, and Lumen Technologies has disrupted NetNut, a residential proxy botnet that had quietly compromised at least two million Android devices — smart TVs, streaming boxes, and phones — according to Bleeping Computer. Infected devices were used to route criminal and espionage traffic through ordinary home internet addresses, making the traffic look legitimate.

Devices typically get infected through pre-installed malware or trojanized apps downloaded outside the official Play Store.

What to do right now:

1. Open the Google Play Store, tap your profile icon, and select Play Protect → Scan. Google has been automatically disabling infected apps for enrolled devices.
2. Avoid sideloading apps (installing from outside the Play Store) on Android TV boxes or streaming sticks.
3. If you use a cheap Android streaming device from an unfamiliar brand, consider replacing it with a device that receives regular security updates.

ClickFix attacks work by tricking you into copying a dangerous command, usually disguised as a verification step or a helpful fix, and then pasting it directly into your computer's terminal. The result is malware running with your own permissions, bypassing most security tools entirely.

Opera has introduced Paste Protect, a feature that blocks suspicious commands before they ever reach your clipboard. When it detects a harmful pattern, it stops the copy action, shows you a warning, and flags a red indicator in the address bar. You can still review the first 120 characters of what was blocked and approve it manually after a short delay if you know it is safe.

The feature is on by default in the latest Opera release. You can manage it at Settings → Privacy & Security → Paste Protect.

Get Opera at opera.com — it is free. Full details on Paste Protect via Bleeping Computer.

A ClickFix attack starts with something completely ordinary: a video that will not load, or a CAPTCHA that keeps failing. A pop-up offers a quick fix and asks you to copy a command and paste it into your computer's terminal. The command installs malware — and you ran it yourself, which means most security software never stops it. Huntress reports this technique now accounts for over 53% of malware-delivery attacks.

Opera has added a feature called Paste Protect to its desktop browser, enabled by default, that monitors clipboard activity in real time and blocks malicious commands before you can paste them. If it detects something suspicious, the action is blocked and a warning appears immediately.

If you use Opera, you are already protected. If you use another browser, the rule is simple: never paste a command into your terminal because a website told you to.

Download Opera here.

Microsoft has rolled out a new admin policy for Microsoft Teams (the video conferencing and collaboration platform) that detects external bots before they join your meetings and holds them in a lobby for organiser approval, according to Help Net Security.

Previously, bots could join meetings with the same one-click admit as any other participant. Under the new policy, Teams uses behavioural and infrastructure signals to identify bots, groups them separately in the lobby, and removes the one-click admit option for them. The organiser sees a clear visual indicator and must explicitly approve entry.

What to do: If your organisation uses Teams, ask your IT admin to enable the Manage external bots and their access to meetings policy in the Teams Admin Center. Also set Who can admit from the lobby to organizers and co-organizers only. If you run meetings yourself and have admin access, this takes about two minutes to configure.

WhatsApp is rolling out usernames globally today, and reserving yours is worth doing immediately, according to The Hacker News.

Once enabled, your phone number becomes invisible to anyone who contacts you via username. They need your exact username to find you — and optionally a username key (a secondary code you set that new contacts must also know before they can message you). No public directory exists, so you cannot be discovered by browsing.

This matters because phone numbers are a common starting point for SIM-swapping attacks, spam, and targeted scams. Removing your number from the equation cuts off one well-travelled route.

To reserve your username: Open WhatsApp, go to Settings > Account > Username, and follow the prompts. The feature is rolling out gradually, so it may not appear immediately.

Mozilla's 0DIN team demonstrated today that malicious code does not need to live in a repository to compromise your machine. The attack exploits the trust developers place in AI coding tools — the agent reads setup instructions and follows them faithfully, with no way to see what will be fetched at runtime.

The practical defence is to review any initialisation or setup command before running it, regardless of whether an AI tool recommends it. Before executing a shell script from an unfamiliar source, check where it fetches content from. If a command pipes anything directly to bash or sh — particularly one that resolves an external address — treat it as suspicious.

0DIN's full write-up, including the proof-of-concept chain, is available via Help Net Security. If you work with developers or manage engineering teams, sharing this finding today is worthwhile.

Fake support messages from Russian intelligence have put messaging account security back in the spotlight, and the fix does not require any special software.

Both Signal and WhatsApp let you see every device your account is currently linked to. If an attacker has used a stolen recovery key or confirmation code to link your account to their device, it will show up here.

On Signal: Go to Settings → Linked Devices. Remove anything unfamiliar.

On WhatsApp: Go to Settings → Linked Devices. Review the list and tap any unrecognised session to remove it.

While you are there, enable two-step verification on both apps. Signal: Settings → Account → Two-Step Verification. WhatsApp: Settings → Account → Two-Step Verification.

This takes under five minutes. The SSU advisory behind this story is available via The Hacker News if you want the full detail on how the campaign operates.

Russian intelligence groups are actively phishing Signal users out of their Backup Recovery Keys, according to an updated FBI and CISA advisory covered by The Hacker News. Hand that key to an attacker and they can restore your entire message history and take over the account. The old key keeps working even after you change devices.

The fix is fast. Open Signal, go to Settings > Linked Devices, and remove anything unfamiliar. Then go to Settings > Account > Backups and generate a new Recovery Key. That kills the old one.

One firm rule: Signal support does not message you inside the app. Any message asking for your Recovery Key, a PIN, or a verification code is an attack. Treat it as one.

This applies to high-profile targets first, but the tactics will broaden. Anyone using Signal for sensitive conversations should do this today.

Security expert Richard Bejtlich has published a free practical guide on network detection and response (NDR) — the discipline of monitoring traffic moving inside your network, not just at its edges — in partnership with Corelight, according to The Hacker News.

The core argument is worth understanding. Alerts tell you something may have happened. Network evidence tells you what actually did. Bejtlich calls this interdiction: catching attackers after they get in but before they achieve their goal. Today's phishing campaign targeting hotels is a perfect example — the implant was already running before any alert fired.

If you manage a small network, the practical takeaway is this: start logging outbound traffic, not just inbound. Unusual connections to unfamiliar IPs on non-standard ports are often the first real sign of compromise. The free guide is available via the link above.

Social engineering (manipulating people through deception rather than breaking through technical defences) is behind some of the most damaging breaches of recent years. In 2025, the hacking group Scattered Spider hit UK retailers Marks & Spencer, Co-op, and Harrods by impersonating employees and convincing service desk agents to reset credentials and disable multi-factor authentication, according to Bleeping Computer.

If you manage or work at an organisation with an IT help desk, one change matters most: require staff to verify identity through a second channel before any password reset or MFA change. A callback to a verified internal number — not the number the caller provides — breaks the most common attack chain entirely.

For individuals: if your workplace IT desk ever contacts you unexpectedly, call them back on a number you looked up yourself, not one they gave you.

GitHub has updated its actions/checkout tool — the official action that pulls repository code into automated build pipelines — to block a class of attacks called pwn requests, according to The Hacker News.

A pwn request attack works like letting a stranger into your office to drop off a delivery, and then discovering they rewired the server room while no one was watching. Attackers submit a pull request containing malicious code; a misconfigured workflow runs it automatically with full access to secrets and tokens.

As of June 18, actions/checkout v7 refuses to check out unreviewed fork code in the most dangerous workflow configurations by default.

What to do: If you maintain any GitHub Actions workflows, update actions/checkout to v7 now. Review any workflows using the pull_request_target trigger and confirm they do not check out code from untrusted forks. The full backport to older supported versions lands July 16.

OpenAI has rolled out an upgraded AI model called GPT-5.5-Cyber through its Daybreak security initiative, and alongside it, an updated Codex Security plugin that goes beyond flagging vulnerabilities to actually generating patches. According to The Hacker News, the tool can scan large codebases, trace attack paths, validate findings, and produce codebase-specific fixes ready for human review.

If you maintain or contribute to open-source software, the associated Patch the Planet initiative — run in partnership with security firm Trail of Bits — is accepting participation from projects including cURL, Python, and others.

Access is currently limited to trusted defenders through the Daybreak programme. If you qualify, it is worth applying. If you do not, watch this space — the tooling is moving fast.

Learn more and apply here

The OXLOADER campaign works because a fake download site looked real enough to click. The entire attack collapses if you arrive at the genuine download page.

Before downloading any software, find the official website independently — not through a sponsored search result. Search results at the top of the page marked "Sponsored" or "Ad" are paid placements. Anyone can buy them, including attackers.

For developer tools or any software you plan to install, go directly to the project's official domain. For Node.js, that is nodejs.org. Bookmark it. Never download from a URL you arrived at via an ad.

This one habit would have stopped every person targeted by REF8372.

A recently disclosed flaw in Gravity SMTP, a popular WordPress email plugin, exposes a site's full system report — including API keys and email service credentials — to anyone who knows the right URL. Wordfence has logged over 17 million exploit attempts against it, which means attackers are scanning broadly, not hand-picking targets.

If you run a WordPress site, you do not need to wait for the next disclosed plugin flaw to check your exposure. Free tools like Wordfence can scan your installed plugins for endpoints that return data without requiring a login — exactly the pattern behind this flaw and many others like it.

What to do: Run a Wordfence scan on your site today, even if you do not use Gravity SMTP. If you do use it, update to 2.1.5 immediately and rotate every API key and OAuth token tied to your email integrations — treat them as already exposed.

A new ransomware operation called Prinz Eugen is quietly active, and it does not leave a ransom note — so you might not know you've been hit until your files are already encrypted and gone. The group gets in through stolen RDP credentials (remote desktop login details), then manually runs their payload and sets up a backdoor admin account for persistence.

Malwarebytes' enterprise arm ThreatDown has published a full analysis including indicators of compromise — specific file names, IP addresses, and behavioural patterns you or your IT team can use to check whether this threat has touched your environment. You can find the indicators of compromise list in ThreatDown's full report.

What to do right now: Make sure RDP (Remote Desktop Protocol — remote access to your computer) is either disabled or protected by multi-factor authentication. Unexpected new administrator accounts are a red flag worth investigating immediately.

A researcher at the SANS Internet Storm Center spotted a phishing email targeting a major Belgian bank that used a sneaky trick: disguising the malicious link as what looks like an IPv6 address (the newer, longer format for internet addresses) to slip past basic security filters, according to SANS ISC.

The trick works because the link format [::ffff:5511:74be] looks like gibberish to a human but actually encodes a normal web address in hex code. Security tools that scan for suspicious domain names often miss it entirely because there's no domain name to catch, just numbers in brackets.

What to do: You don't need to decode anything yourself. Treat any link in an email that looks like a string of numbers and colons in brackets as an instant red flag, especially in messages claiming to be from your bank. Don't click it. Go to your bank's app or type the website address in yourself instead.

If your organization runs a Fortinet or FortiGate VPN, do not wait for IT to mention it. Ask directly: has our VPN password been rotated since the FortiBleed leak?

Two steps matter most. First, every FortiGate VPN password should be reset, since the leaked database reportedly held working passwords for thousands of devices. Second, MFA, a second login step like a code from your phone, should be turned on for VPN access if it is not already. MFA means a stolen password alone is not enough to get in.

If you are not sure whether your company uses FortiGate, ask your IT contact today. This is the kind of leak that turns into a real breach only when nobody bothers to act on it.

The FBI issued a public warning this week about a twist on cryptocurrency investment scams — also known as pig butchering (where fraudsters spend weeks or months building trust with a target before steering them toward a fake investment platform). Scammers are now dispatching physical couriers to collect cash directly from victims after legitimate banks block suspicious wire transfers.

The courier will arrive with an agreed-upon password or a specific banknote serial number to prove they are "from the platform." They are not. The money goes straight to the scammers.

The script is always the same: an online contact — met through dating apps, social media, or messaging platforms — builds a relationship, introduces a lucrative investment opportunity, and then the withdrawals mysteriously stop working. Now there's a new urgency: your account has been "flagged" and you need to hand over cash in person to fix it.

What to do: No legitimate investment platform sends a courier to your door for cash. If anyone online asks you to hand money to a stranger, stop contact immediately and report it to the FBI at ic3.gov.

The FBI, Google, and Black Lotus Labs dismantled Outsider Enterprise this week as part of a broader operation called Operation Riptide, targeting criminal cybercrime infrastructure at scale. During the technical takedown, the agencies dismantled the platform's infrastructure used to distribute the platform's phishing kits to paying customers.

For everyday people: over 9,000 fake websites and more than a million fraudulent URLs are now offline. Campaigns that were impersonating trusted brands via SMS across AT&T, T-Mobile, and Verizon have been severed from their infrastructure.

Nothing you need to do — this one is a win. But if you received a suspicious text in the past few months impersonating a delivery company, bank, or government agency, check your credit card statements for anything unfamiliar.