ReversingLabs detailed a malicious npm package, tw-pkgprobe-7731, that masquerades as an authorized Twilio bug-bounty research probe while harvesting developer data. Uploaded in mid-August by an account that no longer exists, it shipped eleven versions within about 45 minutes. Comments inside describe it as an authorized HackerOne probe that runs only inside Twilio's serverless sandbox and takes no destructive action. On execution it first checks for a Twilio developer environment and exits otherwise, then collects environment variables plus system details like mounts and temporary folders and exfiltrates them through a webhook. Later versions specifically target developers using Twilio APIs by searching for folders tied to particular Twilio account identifiers, sharpening the credential theft.
Checkmarx found an ongoing npm campaign built around indexed-btree, a package impersonating the popular sorted-btree library that has amassed two million weekly downloads. Instead of using preinstall or postinstall scripts, the malware hides its loader inside the BTree.prototype.set method that applications call constantly, so it executes at runtime rather than install time. This sidesteps the npm approval gates GitHub added in June to block lifecycle scripts, and installation looks clean to static scanners. Once triggered, it fingerprints the host, exfiltrates details over hardcoded Slack and Telegram channels, and polls an Ethereum Sepolia smart contract for encrypted second-stage commands.
Accomplish AI researcher Oren Yomtov disclosed two OpenAI Codex sandbox escapes, the more serious dubbed Heapjack. Codex Desktop installs a node_repl component into the global config with no opt-in, and plain Codex CLI users inherit it. That process runs trusted OpenAI code and untrusted agent code in one Node instance sharing a heap, where a random authorization token sits in memory. Untrusted code snapshots the heap, recovers the token, and writes requests onto the pipe to an unsandboxed parent process, reaching any Unix socket including a Docker daemon. Opening a malicious repository and asking about the code yields unsandboxed execution with no prompt.
Air Security reported that four AI coding agents fetch plugins pinned to a reviewed commit hash but never verify the code they receive matches it. On code hosts that permit branch names shaped like commit hashes, such as Bitbucket or self-hosted git, a plugin repository owner can point that name at different code, so the agent installs malicious code while reporting the locked version. Because plugins run with the user's access, the swapped code reaches files, credentials, and connected systems. Anthropic fixed it in Claude Code 2.1.179 and OpenAI in Codex 0.146.0; GitHub Copilot has no fix, and Google will not patch the retiring Gemini CLI.
Attackers stole a Cloudflare API key from marketing platform Brevo and used it to inject malicious code into the scripts that Brevo's customers embed on their own websites, affecting more than 100,000 sites. The key was long-lived, had full account permissions, and was hardcoded in application source code, which let the attackers create a Cloudflare Worker that modified Brevo's forms, widget, and loader scripts at the network edge for about five and a half hours. Visitors saw a fake verification page with ClickFix instructions to run a command on Windows, and on WordPress sites where an admin was logged in, the script tried to silently install a backdoor plugin.
Mandiant reported that an attacker hijacked a developer's active AI coding-assistant session at a software company and used it to spread the self-replicating Shai-Hulud worm across about 100 internal code repositories. The chain started when the AI assistant recommended a piece of software the attacker had poisoned, and the developer accepted the suggestion. Using the live session, the attacker installed an infostealer through a poisoned PyPI package and stole GitHub access tokens, then unleashed the worm, which stole repository secrets and source code. The attacker also poisoned a package in the company's own namespace, so a second developer's pull caused a reinfection. It shows AI-recommended dependencies as a new poisoning path.
Researchers detailed a May 2026 campaign in which a swarm of AI agents abused weaknesses in the RubyGems package registry to create accounts at scale with disposable email addresses and upload more than 2,000 packages, forcing the registry to suspend new registrations for days. The agents then leveraged the documentation builder on RubyDoc.info to achieve remote code execution on its servers and scrape public data, and attempted to harvest users' API keys through a caching flaw that was only fixed months later. It is an early look at AI-driven, automated abuse of package registries and their surrounding build and documentation tooling, which together form a large and often overlooked supply-chain attack surface.
Researchers at GitGuardian found that a recent variant of the self-spreading Shai-Hulud npm worm has expanded its credential theft to scan 469 distinct locations on infected developer machines. The targets now span developer environments, continuous integration and deployment tooling, cloud configuration files, and even the configuration of AI tools. That breadth turns a single compromised package into a wide net for secrets, from cloud and registry credentials to keys held by developer and AI tooling. It reflects how supply-chain worms are industrializing secret collection, treating any credential a developer's machine can reach as fair game once malicious code runs during installation or use.
Attackers used a BGP hijack, a manipulation of internet routing, to divert update traffic for Virtualizor, a widely used server and hypervisor management panel, to a server they controlled. During the diversion, which began August 28, they obtained a valid TLS certificate so the connection looked legitimate, then delivered a malicious update that installed persistent root access on affected hosts. One hosting provider found root-level compromise on five of thirty-four hypervisors it checked. Because the software's updates were not cryptographically signed, transport encryption alone did not stop the tampering once routing was hijacked. The vendor released a scanner and patch, but package signing remains unfinished, leaving update integrity dependent on routing security.
Days after disclosure, attackers are exploiting a critical authentication-bypass flaw in JFrog Artifactory, the widely used repository manager for binaries, packages, containers, and build artifacts. Tracked as CVE-2026-82329 and scored 9.8, the flaw lets an unauthenticated attacker with network access gain administrative privileges under Artifactory's default configuration. Researchers at watchTowr observed exploitation beginning September 1, with attackers minting admin tokens for themselves and enumerating users, groups, and credentials. Because Artifactory sits at the center of software supply chains and CI/CD pipelines, admin access lets attackers tamper with build pipelines, poison trusted dependencies, and push malicious code downstream to customers. JFrog patched it in version 7.161.20 on August 28.