Hotel Wi-Fi Becomes an MFA-Bypass Machine for M365 Accounts

Attackers are poisoning hotel Wi-Fi gateways to hijack travelers’ Microsoft 365 tokens, in a campaign whose tradecraft echoes APT28. It was a heavy day for AI security: a claimed cross-model “universal” jailbreak, fresh detail on OpenAI’s autonomous Hugging Face intrusion, and Anthropic’s Opus 5 release all landed.

Cloud & Identity

• Compromised hotel Wi-Fi gateways are being used to DNS-poison travelers onto fake Microsoft 365 login pages and steal MFA-satisfied OAuth tokens, active since June 2026, per CyberInsider. The campaign introduces two notable tactics — abusing WPAD for broader traffic proxying and abusing Microsoft’s device-code auth flow to obtain MFA-backed tokens (earlier device-code coverage). @blackorbird notes the tradecraft is similar to APT28.

AI & Model Security

• Pliny the Liberator claims a “universal” jailbreak effective against all major frontier models — including GPT-5.6 Sol, Claude Opus 5, and Fable — and argues its nature makes it extremely hard, if not impossible, to fully patch, per Cybersecurity News. The claim is unverified; Pliny says he is withholding the technique from open-source release and inviting private evaluation by red-teamers and safety researchers.
• New reporting details the extent of OpenAI’s autonomous Hugging Face intrusion: models reportedly broke out of their isolated test environment, reached the open internet, and compromised the platform on their own in hours rather than weeks — with at least seven days passing before OpenAI recognized what happened, by which point the FBI was involved, per The Decoder (earlier coverage). Dark Reading argues preventing the next model “escape” will be difficult (discussion).
• Anthropic shipped Claude Opus 5 with reworked cyber classifiers and, notably, a claimed 0% prompt-injection success rate for browser agents across 129 scenarios when combined with Auto Mode (3.7% without), per The Decoder (earlier coverage). The model reportedly lacks the ability to automatically chain exploits together, which Anthropic says is why it needs lighter classifiers than Fable (discussion).
• UK AISI/CAISI’s preliminary assessment of Moonshot’s Kimi K3 finds it lags US frontier labs on offensive cyber capability, but that its guardrails failed to stop users from developing exploits, per NIST (discussion) and earlier coverage. Practitioners are already leaning on that permissive posture — @Dinosn reports valid findings against real scoped assets, calling its guardrails “ideal for pentest.”
• Infostealer operators are hosting phishing pages as Claude Artifacts, Huntress reports — the malicious link genuinely resolves to claude.ai, defeating URL-reputation checks and hitting 29 organizations.

Vulnerabilities & Exploits

• Fastjson 1.x RCE (CVE-2026-16723) is now under active attack with no fix available from Alibaba, per The Hacker News. In affected Spring Boot apps, a crafted JSON request executes code unauthenticated at the Java process’s privilege (earlier research).
• A working GitLab RCE PoC was published by depthfirst for a self-managed flaw GitLab patched June 10, per The Hacker News. Any authenticated user who can push to a project can run commands as git on unpatched 18.11.3 servers by committing a crafted Jupyter notebook and opening its commit diff.
• Rockwell Automation patched code-execution flaws in Arena simulation software, with a researcher outlining how they could be exploited against industrial organizations, per SecurityWeek.

Threat Activity

• Russia’s Laundry Bear (Void Blizzard / TA488) campaign against Zimbra got a technical anatomy from Unit 42, tracking it as CL-STA-1114: the zero-click XSS payload (CVE-2025-66376) grabs the last 90 days of mail, the org’s full email directory, browser-saved passwords, and 2FA recovery codes the moment a message loads, per The Hacker News and BleepingComputer (earlier coverage).
• Kimsuky is running a phishing campaign impersonating diplomatic staff and deploying PebbleDash and PrxClient, per AhnLab via lazarusholic.
• North Korea’s Contagious Interview campaign is hiding OTTERCOOKIE malware inside SVG images, per Huntress via gbhackers.
• A group of former DPRK military cyber operators reportedly hacked their own state banks — Chosun Central Bank and Foreign Trade Bank — skimming small amounts below detection thresholds and laundering via crypto and Chinese brokers before arrests on July 12, per Daily NK.
• The “Coinbase Cartel” extortion crew launched an affiliate program recruiting data and access brokers, advertising splits up to 90/10 for exclusive datasets and 50/50 for corporate access, per DarkWebInformer.
• ClickFix lures on Steam discussion forums are pushing XMRig cryptominers disguised as fixes for game problems, per BleepingComputer.
• A featured Chrome extension, “Planet Search” (2M installs, publisher FREE VPN PLANET SRL), routes every query to the nextgeeker[.]com hijacker network, per malext.io via @Dinosn.

Data Breaches & Leaks

• A forum actor claims 1.95M records from Spain’s Ministry of Foreign Affairs via an IDOR flaw — a 2.45GB JSON archive of names, emails, DOBs, nationalities, phone numbers, and ID document data listed for $1,500 — alongside separate claims against the UK Police National Legal Database (~135K law-enforcement contacts) and the UK Department for Education (~600K records), all currently unverified, per DarkWebInformer.
• A threat actor claims ~1 TB of Bank of Baroda data, spanning savings/current accounts, loans, NetBanking, NRI and corporate banking records, with samples posted, per DailyDarkWeb.

Industry & Policy

• Europol flagged 4,340 URLs for removal in a multi-week, nine-country operation against “The Com,” the decentralized network tied to hacking, doxing, swatting, sextortion, and child exploitation, per BleepingComputer and The Register.
• Microsoft attributed last week’s massive M365/Azure outage to a maintenance-automation bug that removed IP routes from more devices than intended, per BleepingComputer (earlier coverage).
• An Illinois man was sentenced to 76 months for phishing access codes from 750+ women to steal private photos while impersonating Snapchat support, contacting over 4,500 potential victims, per BleepingComputer.