400 Arch Linux Packages Were a Trap

The Hacker News reports that attackers compromised more than 400 packages inside the Arch User Repository (AUR) — Arch Linux's community-run collection of installable software, maintained separately from its official repositories. The official repositories were not affected.

The attackers did not find a flaw in the software itself. They targeted the package maintainers — the trusted individuals responsible for keeping each package current — and once inside their accounts, rewrote the build scripts (the step-by-step instructions your machine follows when installing software) to deliver malicious code alongside the legitimate program.

That malicious code is a Rust-compiled credential stealer: a compact, hard-to-detect program built to harvest developer secrets like SSH keys, API tokens, and stored passwords. On machines where it gained elevated (root) access, it could also load an eBPF rootkit (a hidden program embedded deep in the operating system's core, extremely difficult to find or remove once installed).

The packages kept their names, histories, and reputations intact. Only the build instructions changed. The trap looked exactly like software developers already trusted.

The affected-package list is still growing.

What to do: If you installed or updated any AUR package on or after June 11, check it against the current affected list before trusting that machine.

Google is taking legal action against a Chinese cybercrime network accused of building Outsider, a phishing-as-a-service (PhaaS) platform — a packaged criminal toolkit that lets others run phishing campaigns without building their own infrastructure. The network weaponised Google's own Gemini AI agent to generate convincing fraudulent web pages, then deployed them through mass SMS campaigns targeting Americans.

Messages mimicked banks and mobile carriers, directing recipients to counterfeit sites built to steal personal and financial data.

Google is now partnering with AT&T, T-Mobile, and Verizon to block these messages at the carrier level and is pursuing the lawsuit to dismantle the operation's infrastructure.

This is a confirmed case of a commercial AI model being turned against the users of the very company that built it.

What phpBB is: phpBB is a free, open-source web forum platform still running thousands of active community discussion boards worldwide.

What it is: A flaw in phpBB's authentication logic (the system that checks whether you are who you say you are) allows an attacker to skip the login process and access any account on the forum, including administrator accounts. Researchers at Aikido discovered it on June 2 and reported it immediately. phpBB patched it four days later in version 3.3.17.

Who's at risk: Anyone running phpBB 3.3.16 or below, or the 4.x alpha 4.0.0-a2. The 4.x branch has no patch available yet.

Google filed a lawsuit this week against a Chinese cybercrime network running mass smishing (smishing: phishing via text message) campaigns against Americans. The texts impersonated banks and phone carriers, claimed urgent account problems or uncollected rewards, and linked to fake websites built to steal personal and financial information.

The defence costs nothing: do not click links in unexpected texts, regardless of how legitimate the sender appears. If the message sounds urgent, open the company's official app directly or type their website address yourself. Never follow the link in the text.

Your phone carrier almost certainly has a free spam-blocking feature. Check your settings or contact them to switch it on.

On June 12, the U.S. government issued an export control directive ordering Anthropic to suspend all access to Fable 5 and Mythos 5, its two most advanced AI models, for every foreign national worldwide, including Anthropic's own foreign-national employees. Because selectively filtering foreign nationals was not technically feasible, Anthropic suspended both models globally.

The stated concern: someone had found a jailbreak — a technique that manipulates an AI model into ignoring its own safety rules, unlocking capabilities the model was specifically built to block. Anthropic is not disputing the jailbreak exists — it argues the technique is narrow and the underlying capabilities are already available elsewhere. All other Anthropic models remain unaffected.

This marks the most significant AI-specific export control action by the U.S. government to date.