
This week in brief
Last week the cyber world invaded the physical world, everyday software was exploited, and election results could determine the future of AI regulation in the US. As always, check out the “What this means for your organization” section for actions you can take now.
The week’s biggest story was the hijacking of internet-connected refrigeration systems. Attackers adjusted supermarket freezers across Europe and French blood bank cold rooms to dangerous temperatures, and US agencies issued a joint warning. It is a reminder that internet connected controllers running building systems can be as exposed as laptops.
Within hours, attackers moved on flaws uncovered in WordPress and OpenSSH, two pieces of software that run a large share of the internet. If you have a website or a Linux server, this is a patch week.
In the US, Democrats won both chambers of Congress, and voters said AI and data centers were a major factor. Meanwhile, the Justice Department launched a “Super Intelligence” task force, and Google, Meta and TikTok went to court to block California’s new limits on addictive feeds for kids.
Beyond the US, the European Commission preliminarily found TikTok’s “For You” feed in breach of the Digital Services Act, and China and South Korea issued new rules and guidance for AI agents. Binance’s incident report also showed how a fake job interview and an internal AI tool opened the door to the largest hack in history.
– Charles Hill
Top 5 developments
- Hackers hijack refrigeration systems across Europe. Attackers changed temperature and defrost settings on internet-exposed controllers at supermarkets in several European countries and at three French blood storage sites. On November 5, CISA, the FBI, HHS, USDA and DOT issued a joint advisory urging operators to take these systems off the internet, BleepingComputer reported.
- WordPress and OpenSSH flaws exploited almost immediately. CISA added a WordPress Core flaw (CVE-2026-119977) to its Known Exploited Vulnerabilities catalog after attacks during core upgrades. Separately, the OpenSSH “SSHole” login bypass (CVE-2026-123158) was exploited within nine hours of disclosure on Alpine-style Linux systems.
- Democrats take Congress, with AI on the ballot. Democrats won the House and Senate on November 3, and exit polling found 56% of voters called AI and data centers a crucial or major factor. Incoming leaders are promising hearings and “reasonable guardrails”, while local data center restrictions passed in Ohio and elsewhere.
- Kids’ online safety fights on both sides of the Atlantic. The European Commission preliminarily found TikTok’s “For You” feed in breach of the DSA over harmful “rabbit holes.” In the US, Google, Meta and TikTok sued to block California’s law limiting addictive feeds for users under 16.
- Binance traces its $6.3 billion hack to a fake job interview. Binance’s incident report says attackers spent months inside employee machines after a fake recruiter delivered malware, then used an internal AI tool to reach wallet developers’ systems before draining hot wallets on October 29.
United States
Democrats win both chambers. The new majorities plan public hearings on AI, and several incoming senators ran on data center moratoria, energy-cost rules or youth safety. They lack a veto-proof majority, so federal preemption of state AI laws becomes harder but new federal AI rules are not guaranteed. Locally, Euronews reported that 15 Ohio municipalities passed data center bans or restrictions and Texas paused new data center permits pending an audit. In a 60 Minutes interview on November 8, President Trump said AI risks are manageable and he is open to guardrails, as long as the US doesn’t fall behind China.
DOJ launches a Super Intelligence Task Force. The Justice Department created the task force to pursue foreign state actors, criminal groups and fraudsters who misuse advanced AI. It reportedly absorbs the AI Task Force announced in January. Expect more prosecutions that cite AI as an aggravating factor in fraud cases.
Tech giants sue over California’s kids’ feed law. Google, Meta and TikTok filed separate federal suits against a law taking effect January 1, 2027. The law bars addictive feeds, endless scrolling and autoplay for users under 16 and requires reasonable age-verification measures. The companies argue it violates the First Amendment.
White House network used as a proxy exit. Researchers at Layer3 Intel traced residential proxy traffic from 28 proxy services to six White House IP addresses over most of October. They found no compromise of White House systems and suspect one infected personal device with a proxy SDK. It shows how a single unmanaged phone or TV box can lend your network’s reputation to criminals.
European Union
TikTok’s feed preliminarily found in breach of the DSA. The Commission said TikTok’s risk assessment and mitigation fell short on repeated exposure to content about sadness, isolation and body image, especially for minors and vulnerable adults. This is not a final decision or fine. Any platform with a recommender system should note that “rabbit hole” design is now an enforcement theory.
Advocate General: AI chatbots may infringe press publishers’ rights. In Like Company v Google, Advocate General Szpunar advised the EU Court of Justice that a chatbot reusing press articles in real time may infringe publishers’ rights in some cases. That depends on whether content was accessed unlawfully (for example, around a paywall) and whether the answer replaces the original. The opinion isn’t binding, but it is the first serious signal on AI answers and EU copyright.
EDPB launches a children’s privacy hub. The EDPB launched “Privacy for Kids”, a set of games, videos and guides from DPAs across Europe, filterable by age and topic. The Commission also opened a call for evidence on DSA advertising repositories. Feedback on the draft AI Act enforcement procedures for general-purpose AI closes today, November 9.
United Kingdom
NCSC warns of AI-powered attacks on critical infrastructure. The NCSC’s State of the Cyber Nation report says AI-enabled malware is highly likely to be used in a ransomware attack on UK critical national infrastructure by 2028. CEO Richard Horne urged organizations to stop buying “tech on the cheap.” The government also announced funding for a national Cyber Shield and a real-time vulnerability center.
Rules for police facial recognition, and the AI Act in Northern Ireland. The government proposed a legal framework and an independent watchdog for police use of facial recognition. It also published guidance on how parts of the EU AI Act apply in Northern Ireland under the Windsor Framework from November 10. There is still no government AI Bill.
Asia-Pacific & Global
China sets rules for AI agents on devices. China’s MIIT issued Notice No. 42, requiring compatibility verification, fraud-risk assessment and anti-fraud testing before phones, tablets and wearables with pre-installed AI agents can access networks. At the Wuzhen summit, the CAC and GSMA announced an AI agent self-discipline convention. The Straits Times reported that Singapore is exploring DNS-based identification of AI agents, a sign that “agent identity” is becoming a policy topic.
South Korea maps out its AI law agenda. The Presidential Council on National AI Strategy approved a “Korea AI Legislative Framework” that prioritizes the AI Basic Act and sector-specific amendments. KISA also published an Agentic AI Zero Trust Guide, a practical reference for anyone deploying agents with access to internal systems.
Ethics & Accountability
Cold-chain attacks cross into the physical world. The French Blood Service confirmed that settings were altered at three blood storage facilities, with some cold rooms pushed far above safe temperatures; the Health Ministry said early detection prevented any impact on supplies. A November wave hit more than 115 stores in six European countries after October attacks on French supermarkets. The Commission issued a joint statement on the incidents, and Censys reported a 53% drop in exposed controllers as operators pulled them offline. No attacker has been named.
WordPress and OpenSSH under active attack. The WordPress flaw is a race condition during full core upgrades that can let an unauthenticated attacker run code; Wordfence logged thousands of attack attempts a day, The Hacker News reported. SSHole lets attackers log in to accounts without a usable password on musl-based Linux without PAM, including Alpine’s default setup. It is fixed in OpenSSH 10.6p2, and Alpine shipped patcheson November 5. Glibc-based systems are not affected.
Binance: a fake recruiter, then an AI tool. Binance’s report says a June compromise of a developer tool and a September fake interview, including a video call with a fake CEO, led to malware on an engineer’s machine. Attackers then used an internal AI tool to take over two wallet developers’ systems. About $2.85 billion had been frozen or intercepted by November 3, and Binance says it has rebuilt the affected AI tool.
AI coding agents install abandoned software. Glow Labs’ “MCPhantom” research found AI coding agents installing deprecated MCP servers that will never be patched, usually choosing them from training data rather than user instructions. The packages include high-severity flaws and have more than 17 million npm downloads. The pattern appeared across tools from Anthropic, OpenAI, Cursor and xAI.
What this means for your organization
- Update WordPress and your Linux servers now. Don’t wait. Both the WordPress Core flaw and OpenSSH’s SSHole are being exploited. Ask your web host or developer to confirm WordPress is on the latest release, and ask whoever runs your servers whether any use Alpine or other musl-based Linux. If so, update to OpenSSH 10.6p2, restart the SSH service, and turn off password logins.
- Take building and equipment controllers off the internet. The refrigeration attacks worked because controllers were reachable online with weak passwords. Ask your facilities, HVAC or refrigeration vendor whether any of your systems can be reached from the internet, change default passwords, and require a VPN or secure gateway for remote access.
- Treat job interviews as an attack path. Binance was breached through a fake recruiter and a “coding test.” Tell staff never to run code or install software for an outside interview on a work device, and verify any unexpected video call from an executive through a separate channel.
- Check what your AI tools can reach and install. An internal AI tool helped the Binance attackers, and coding agents are installing abandoned packages. Write down which systems each AI tool or agent can access, remove access it doesn’t need, and ask developers to review any package an agent adds before it goes live.
- Clean up unmanaged devices on your network. One infected device turned the White House network into a proxy exit. Put personal phones, smart TVs and streaming boxes on a separate guest network, away from business systems.
- Review features that minors could use. The EU’s TikTok finding and California’s under-16 feed law (in effect January 1 unless blocked) both target engagement-driven design. If your app or site has feeds, autoplay or recommendations, note who your youngest users might be and what safeguards you have.
- Plan for more AI rules from Washington, not fewer. With new congressional majorities promising hearings, state AI laws are less likely to be preempted soon. Keep your Colorado, California and New York compliance plans for January 1 on track.
The week ahead & watchlist
| Date | Item |
|---|---|
| Nov 9, 2026 | Feedback closes on the EU draft implementing act for GPAI enforcement proceedings (AI Act Arts. 75c–75d) |
| Nov 10, 2026 | Microsoft Patch Tuesday; partial application of the EU AI Act in Northern Ireland begins; Korea PIPC AI privacy consultation closes |
| Nov 11, 2026 | CISA deadline for federal agencies to fix the VMware ESX flaw (CVE-2026-41709); EU Digital Fairness Act proposal (reported) |
| Nov 13, 2026 | Comments close on EDPB Guidelines 04/2026 on fines; Character.AI sale objection deadline (reported) |
| Nov 15, 2026 | Comments close on China’s draft AI export-control catalogue |
| Nov 18, 2026 | EU AI Board meeting |
| Nov 20, 2026 | ICO agentic AI call for evidence closes; Character.AI bid deadline (reported) |
| Nov 24, 2026 (proposed) | Character.AI sale hearing |
| Nov 26, 2026 | Lords IRDC evidence deadline on multilateral AI governance |
| Late Nov 2026 | France’s public-sector AI procurement guide, part 2 (contract clauses) |
| Nov 30, 2026 | Australian parliamentary AI inquiry final report |
| Dec 2, 2026 | EU AI Act Art. 50(2) marking deadline for existing systems |
| ~Dec 19, 2026 | 60-day window closes for agency exception requests under the DHS DeepSeek order |
| Dec 31, 2026 | NY Gov. Hochul’s deadline on pending AI bills; Colorado ADMT and chatbot rules must be adopted |
| Jan 1, 2027 | Colorado ADMT Act and NY RAISE Act duties begin; NY AI companion law and California under-16 feed law take effect (unless enjoined); many California AI laws take effect |
| Jan 3, 2027 | New Congress convenes with Democratic majorities |
| ~Jan 26, 2027 | NY baby monitor security law takes effect |
| TBC | European Parliament position on the Digital Omnibus (GDPR/data), then trilogues; CJEU ruling in C-250/25 |
Want help applying this to your business?
We help small businesses and startups put practical data and AI governance in place. Start with a complimentary 30-minute consultation.
This briefing is for general information only and is not legal advice. Regulations change often, so confirm current requirements before acting.
