GitHub will change npm's defaults so the install command no longer runs scripts automatically, disabling a feature commonly exploited by malicious packages such as the notorious Shai-Hulud worm. Maintainer Leo Balter said: "Install-time lifecycle scripts are the single largest code-execution surface in the npm ecosystem. Every npm install runs scripts from every transitive dependency, so a single compromised package anywhere in your tree can execute arbitrary code on a developer machine or CI (continuous integration) runner." In npm 12, due July, three security-focused defaults are changing. Scripts configured for preinstall, install, or postinstall will no longer run unless explicitly permitted via allow-scripts. The --allow-git flag, which pulls dependencies from remote URLs, will default to off, closing an attack path where a malicious .npmrc file could override the Git executable and achieve arbitrary code execution. Finally, allow-remote will default to none, blocking dependency downloads from remote URLs entirely. It will still be possible to allow scripts to run via an allowlist in the package.json configuration file. This will be pinned to the installed version of a package by default. These are breaking changes, and Balter recommended developers run the commands to allow scripts for every currently installed package in a project that requires them. "This gets you protected against new, unexpected scripts immediately," he said. The next step is to review these packages and deny scripts for those where they are not needed. Some packages require script approval to function, including native modules that compile on install, testing tools like Playwright and Puppeteer (which fetch binaries via postinstall), and Electron, which wraps the Chromium browser engine for cross-platform desktop applications. These features have been available since npm version 11.10.0, released in February, but as opt-in flags rather than defaults. That version also introduced min-release-age, which blocks installation of package version newer than a specified number of days, designed as a safeguard against newly published malicious packages. Best security practice for developers using npm 11.16, the current version, is to set these flags on in .npmrc or via environment variables, which will also prepare a project for the changes in version 12. One annoyance is that the existing flag ignore-scripts does not support an allowlist, other than via an additional tool. The ignore-scripts setting will override allow-scripts, so developers will need to remove it, if set to true, to enable approved scripts to run. The allowScripts setting exists in npm 11 but is advisory only. Will this fix npm security issues? Unfortunately not. "Now all the malware can move from the install script to the module itself where it will inevitably still be run," said one developer. Another common view is that developers should use pnpm, which already has safer defaults than npm, including a minimum release age. There is consensus, though, that these changes do improve npm security and are long overdue. The pull request for this change includes the remark that "npm is the only remaining major package manager that runs dependency install scripts by default. pnpm v10+, Yarn Berry, Bun, and Deno all block them." ®

It's patch time for Ivanti customers again after the security shop disclosed another two critical vulnerabilities in one of its products. Both bugs affect Ivanti Sentry, a mobile gateway that forms part of its broader unified endpoint management platform. The first and worst of the two is CVE-2026-10520 (10.0), a max-severity vulnerability that allows a remote, unauthenticated attacker to execute code with root privileges. Flaws that allow root-level code execution without authentication are about as bad as vulnerabilities get, which explains the perfect-10 rating. The only saving grace is that, by the vendor's reckoning, no one has successfully exploited it in the wild… yet. Public disclosures tend to start a figurative countdown timer when it comes to attackers exploiting bugs, and although Ivanti gave little away about CVE-2026-10520 in its advisory, other researchers have already published breakdowns of the patch, offering clues as to how unpatched systems could still be attacked. According to watchTowr, the vulnerability stemmed from an exposed API running under Apache Tomcat. An attacker could feed the API a specially crafted message, which is parsed as a MICS configuration command and executed by the backend handler with root privileges. It looks like Ivanti fixed this by preventing this attacker-supplied string from being accepted, replacing it with a single, hard-coded command. It also updated the Apache configuration rules to block unauthenticated access to the affected endpoint. The second critical Ivanti Sentry vulnerability is tracked as CVE-2026-10523, and is scarcely less serious, carrying a near-maximum 9.9 CVSS. The authentication bypass bug allows remote, unauthenticated attackers to create admin accounts, granting themselves top privileges on an affected system. Customers are advised to address both security flaws immediately. They can upgrade to versions 10.5.2, 10.6.2, or 10.7.1. Ivanti's disclosure this week comes after it fixed two separate critical vulnerabilities affecting its Endpoint Manager Mobile (EPMM) in January. The bugs were both handed 9.8 CVSS scores and were exploited as zero-days. Even the Dutch data protection authority reported itself to parliament after attackers breached it as part of the pre-patch exploits. ®

Microsoft set a record with its June Patch Tuesday release, addressing 206 CVEs across its products and shipping fixes for them, with 38 deemed critical and the rest important. Three are listed as publicly known, but none (so far) have been exploited in the wild. We have no idea how many of these June bugs were uncovered using AI tools. Unlike last month’s patching event, when Redmond disclosed its agentic bug-hunting system found 16 of the 137 vulnerabilities, there’s no word on any AI assists for new releases. Still, it’s safe to assume AI played a major role. As Tom Gallagher, VP of engineering at Microsoft Security Response Center, said about May's Patch Tuesday with a whopping 30 critical flaws: “We expect releases to continue trending larger for some time.” June’s Patch Tuesday proved Gallagher correct, surpassing May in both overall volume and critical bugs. “I’ve been counting CVEs on Patch Tuesday since 2017, and this is by far the largest monthly release in that time,” Zero Day Initiative’s bug hunter in chief Dustin Childs said in his review. “It is extraordinary that Microsoft can produce so many patches in a single month, but it does raise concerns,” he added, asking, as we did: How many were found via AI? And: “How many patches were generated using AI to assist in coding or testing? What quality issues may exist in these patches? And likely most importantly, is this the new normal?” Childs noted that May and April also saw mega releases. “Should sysadmins adjust their processes for prioritization and patch deployment based on this new volume of updates? Unfortunately, Microsoft is not providing those answers right now,” he wrote, adding in this fun fact: “The current number of CVEs shipped by Microsoft this year exceeds the total number of CVEs shipped in all of 2018.” Wowza. While it’s fun to watch from a purely speculative standpoint, as in: "Will Microsoft top 300 next month?", our thoughts and prayers are nonetheless with sysadmins and vulnerability management teams drowning in the AI-induced vulnpocalypse by now. None of the Patch Tuesday security holes are listed as under attack – at least not yet – but three are listed as publicly known. Let’s take a look at those first. Three known vulnerabilities CVE-2026-49160 is an HTTP.sys denial of service vulnerability that we wrote about earlier this month. Calif researcher Quang Luong discovered the attack with an assist from OpenAI's Codex agent, named it HTTP/2 Bomb, and said it exploits the HTTP/2 header compression algorithm by sending thousands of tiny messages to the server, forcing it to rapidly allocate memory and ultimately crash. At the time, a Microsoft spokesperson told The Register that Redmond was “aware and actively investigating appropriate mitigations.” On Tuesday, the tech giant fixed the security issue by introducing a new MaxHeadersCount registry setting, which allows users to limit the number of headers included in HTTP/2 and HTTP/3 requests, and should prevent denial-of-service attacks. CVE-2026-50507, a security feature bypass bug in Windows BitLocker, is the second CVE listed as publicly disclosed, and “exploitation more likely.” An attacker with physical access to the vulnerable system could bypass the BitLocker Device Encryption feature and gain access to the device's encrypted data, according to the advisory. This flaw also seems to be a patch for one of the zero-days dropped in the ongoing war between Microsoft and a disgruntled bug hunter known as Nightmare Eclipse - likely the YellowKey vulnerability disclosed in May. Nightmare has published details about and in some cases, full proof-of-concept exploit code for six zero-days, and promised a “bone shattering” release on June 14. The third publicly known bug, CVE-2026-45586, is a Windows Collaborative Translation Framework (CTFMON) elevation of privilege vulnerability that can be abused by an authorized attacker to elevate privileges locally and gain SYSTEM access. From there, miscreants could deploy malware, steal data, and move laterally through the victim's environment - so patch this one sooner. Plus these two (of 38) critical bugs In addition to those three known vulnerabilities that made the rounds before Microsoft issued a patch, a couple of critical-rated 9.8 security flaws are worth highlighting this month. The first, CVE-2026-45657, is a Windows kernel remote code execution (RCE) bug that allows remote, unauthenticated attackers to run code with system-level privileges without any user interaction. It’s due to an error in how the Windows kernel processes some TCP/IP data, and can be exploited by sending malicious network packets to a vulnerable Windows system, thus triggering the flaw. While it’s listed as “exploitation less likely” by Redmond, we like Childs’ response. “Rest assured that every researcher and bug shop on the planet is reversing this patch right now trying to create an exploit,” he said. “Test and deploy this patch quickly.” CVE-2026-47291, an HTTP.sys RCE vulnerability that also earned a 9.8 CVSS rating, deserves attention as it can also be triggered with zero user interaction and Microsoft says it’s “more likely” to be exploited. “This vulnerability creates severe business risk because HTTP.sys is used by Windows services that process HTTP traffic,” Alex Vovk, CEO and co-founder of patch-management vendor Action1, told The Register. “A successful attack could lead to server takeover, malware deployment, data theft, service disruption, and lateral movement across the environment. Internet-facing systems are especially exposed.” The good news: systems using the Windows HTTP stack’s default MaxRequestBytes registry value are not affected. In the advisory, Redmond provides detailed instructions on how to edit registry settings, which can buy admins some time (and security) while deploying the patch. ®

As if the Miasma situation weren't bad enough, now this weapon is spreading like wildfire. Someone open sourced the entire Miasma worm supply-chain attack toolkit, likely using previously compromised developers' accounts to publish GitHub repositories containing the self-spreading malware’s source code over the last 24 hours. SafeDep, a company focused on open source supply chain security that developed Package Management Guard (PMG), spotted the malicious repos, named “Miasma-Open-Source-Release,” and said that they started appearing on Monday. Its researchers analyzed one of these before GitHub nixed it, and described the code as more than just a supply chain worm. “It is a full supply chain attack toolkit that allows the operator to execute various attacks via stolen credentials against arbitrary or targeted packages on public registries (PyPI, npm, RubyGems), JFrog Artifactory, GitHub repositories and GitHub Actions, AI coding tools config poisoning, SSH based lateral movement and other attack vectors,” the SafeDep team said. While we don’t know who is behind this publicly released worm, it follows in the footsteps of TeamPCP, which developed and then open sourced the mini Shai-Hulud worm last month, announcing a supply-chain attack contest on BreachForums and spawning copycat open source package poisonings. One of these copycat worms, Miasma, first hit upwards of 100 Red Hat and Microsoft open source projects before spreading to other victims, with app-security firm Socket tracking 473 affected package artifacts as of Tuesday. “The Miasma repository is an evolution of the Mini Shai-Hulud toolkit, and was open-sourced June 8 via four previously compromised users,” Rami McCarthy, principal threat researcher at Wiz, told The Register. “Since we had already reversed the payload, this public release isn’t particularly useful for sophisticated defenders, and we haven't observed any opportunistic adoption of it yet.” This, he added, mimics what happened when TeamPCP open sourced mini Shai-Hulud last month. “We didn't see attackers weaponize it either,” McCarthy said. “It's not clear [whether] attackers benefit from adopting this out-of-the-box toolkit versus vibe coding their own. And while it raises concerns about muddying attribution, attackers tend to continue developing their private fork of the malware, providing a clear payload progression to track and deconflict from anyone utilizing the open-source version.” An interesting aspect of both of these worms and other recent attacks like this one dubbed “Comment-and-Control” by AI bug hunter Aonan Guan is that they run entirely in GitHub - they don’t require any custom command-and-control (C2) infrastructure - and use the code-hosting platform for all stages of the attack including remote command execution, configuration, and data exfiltration. “This is a key behavioural shift because traditional network based detection and protection tools rely on baselining and anomaly detection,” SafeDep researchers noted. “Defenders now have to operate closer to application protocol to identify behavioural anomaly instead of network based anomalies.” The Miasma worm uses three independent GitHub commit search channels for C2, and each has a different search string and purpose. One of these, "DontRevokeOrItGoesBoom," discovers attacker-controlled personal access tokens (PATs) to exfiltrate credentials and other sensitive data. These PATs are AES-256-CBC encrypted in the commit message. The second, "TheBeautifulSandsOfTime," delivers JavaScript for immediate command execution. It’s checked once at startup, and, after validation, it passes the payload to eval() to execute at runtime. Finally, “firedalazer” delivers Python script URLs for the persistent monitor. All three are unauthenticated by default, use GitHub’s public commit search API, and use a different validation or decryption key, which means compromising one doesn’t automatically compromise the other two.®

Apple says that its next-gen operating system will allow users to update their weak and compromised passwords with a single tap. Upgrades coming to iOS 27, announced at Tim Cook’s last Worldwide Developers Conference (WWDC) this week, introduce a significant change to the way users manage their passwords. “Building on its ability to alert users about weak and compromised passwords, Passwords can now automatically fix these for users with just a tap,” Apple said on Monday. “Using Apple Intelligence and Safari to agentically take action on a user’s behalf, Passwords securely navigates through websites to sign in and upgrade their accounts to strong passwords.” The iGadget-maker’s existing password manager already flags passwords that are known to be included in prior data breaches, checking whether they appear in known data leaks. However, current Passwords still requires users to update affected accounts themselves and does not offer a way to change multiple compromised credentials at once. Selecting one of those alerts typically takes users to the relevant account page, where they must complete the password change manually. The new update is designed to remove much of that legwork, with iOS 27 automatically navigating supported websites and updating eligible accounts to stronger passwords after user approval. Of course, in the very brief section of the video in which the new capability was announced, the feature worked flawlessly. In practice, however, it remains to be seen how effective Passwords is at agentically navigating different websites’ login processes on behalf of users, especially if MFA is also set up on the account. And for those of you who remember a story The Register covered earlier this year about the (in)security of AI-generated passwords, fret not. Apple’s Passwords app generates solid passwords by default – strings that, according to NordPass’ online password checker, are “strong” and would take centuries to crack. Security company Irregular’s research from February looked at scenarios where users were querying LLM chatbots for password ideas, rather than looking at those generated by purpose-built password managers. Siri state of affairs As predicted by many, this year’s WWDC put Siri, now known as Siri AI, front and center as Apple looks to deliver on its promises made two years ago. It announced Apple Intelligence in 2024, but the offering has underdelivered on pretty much every count. Analysts who spoke to The Register after the event on Monday were optimistic about what they saw on the AI front, but described Apple’s ability to deliver value for developers and users on its second roll of the dice as a credibility test. The company announced a wide range of small AI-enabled upgrades coming soon to iOS 27, powered by Apple's Foundation Models, developed in collaboration with Google and its Gemini technology, in addition to the agentic password-fixing tease. Individually, these features, such as enabling users to create shortcuts or Safari extensions by prompting Apple Intelligence using natural language, and Safari’s Notify Me, which allows users to monitor specific web pages for updates, are not revolutionary. They’re also not the type of features that are poised to set the AI industry alight. But for some, winning the AI race is less about being first to market with the biggest, baddest model; it’s about using AI in the most useful way. "Rebuilt from the ground up, Apple is trying to make AI feel native, useful, and invisible across the devices people already use every day," said Francisco Jeronimo, IDC VP of client devices. "This matters because the winning AI experience for consumers will not be the loudest or most technically complex. It will be the one that understands context, respects privacy, works reliably across apps, and reduces friction without forcing users to change behaviour." Apple’s iOS 27 will launch to the wider public in the fall, while devs can get their hands on the beta version now. This won’t come with the new dedicated Siri AI app, though. You’ll have to join a waiting list for that one. ®