Start Up No.2787: Optimizely the malware spreader, FBI warns of global hacking threat, how the Huangs equipped China, and more


The weird world of the “sovereign citizens” often reflects a desire to escape debt and legal problems. CC-licensed photo by on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.


A selection of 9 links for you. Judicious. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.


Optimizely hack turns world’s biggest websites into malware dispensers • The Register

Carly Page:

»

Some of the world’s most popular websites spent Monday serving malware to visitors after crooks found a way into a piece of software used to A/B test web pages.

Scott Helme, a security researcher at Report URI, identified malicious JavaScript being served by websites belonging to at least 178 organizations, including the BBC, CNN, Zoom, IKEA, Salesforce, HSB, and TalkTalk. All have one thing in common: they use Optimizely’s Web Experimentation software.

Optimizely allows businesses to tweak the content and layout of their websites without changing the underlying site code. Unfortunately, someone other than its paying customers figured out how to do just that.

The attackers abused a feature that allows Optimizely customers to add their own JavaScript to web pages. The malicious code was distributed alongside the legitimate stuff from Optimizely’s own content delivery network, giving customers little reason to suspect anything was amiss.

Optimizely has since confirmed that the criminals exploited a vulnerability, dubbed “OPTIMJACK” by researchers, in the sign-up process for its experimentation software, adding that it had patched the hole, removed the offending code, revoked tokens, and notified affected customers.

…Of the 1,611 Optimizely scripts examined by Helme and his team, 188 contained malicious code, including those used by Optimizely’s own website and login portal. Given the researchers examined only a subset of the scripts, the number of affected businesses could be higher.

The malware underwent several revisions over the course of Monday, with cryptocurrency theft among the crooks’ money-making schemes. Copy a wallet address to your clipboard and the malware could quietly swap it out for one belonging to the attackers, diverting any funds sent to that address.

A card skimmer added later in the day grabbed card numbers, expiration dates, and security codes from payment forms. However, according to a separate analysis by StepSecurity, card details entered into an iframe hosted by a third-party payment processor were beyond the malware’s reach.

«

Small mercies. Exploitation of third-party Javascript, whether here or in malicious ads, is a huge weakness.
unique link to this extract


FBI warns hackers weaponizing ARTEX AI framework to automate attacks on energy, manufacturing, telecom • Industrial Cyber

Anna Ribeiro:

»

The Federal Bureau of Investigation (FBI) warned in a FLASH advisory that hackers are using ARTEX, an open-source, large language model (LLM)-driven autonomous penetration testing framework, to automate multiple stages of cyber intrusions, from reconnaissance and initial access to privilege escalation and data exfiltration. The agency said malicious cyber actors have used the framework since at least August to target organizations across financial services, government, technology, defense, semiconductors, energy, healthcare, manufacturing, telecommunications and legal sectors. Some intrusions progressed from initial targeting to privileged access in less than three hours. 

The FBI highlighted campaigns targeting South Korean financial institutions that resulted in the theft of personal information belonging to approximately 67,000 individuals, as well as a supply chain compromise affecting ConnectWise Cloud Backup customers and enabling access to more than 2,250 downstream tenants. The framework uses specialized AI agents to translate high-level objectives into machine-executable commands, requiring minimal human intervention beyond initial instructions. 

ARTEX is an autonomous, multi-agent penetration testing platform written in Go. It translates high-level tactical goals, described in plain natural language, into machine commands carried out by specialized super intelligence (SI) agents, so little human involvement is needed beyond setting the initial objective. Although its original developer removed the GitHub repository in October following reports of malicious use, ARTEX remains accessible through community forks, archived copies, localized versions and a commercial offering. 

Researchers identified more than 1,600 publicly exposed ARTEX console instances in early November, with deployments likely higher when private and firewall-protected systems are included. 

«

Not proven that this is the cause of any of the recent outages in the US, but it might not be a long time before it is. Open source is forever, of course.
unique link to this extract


Exclusive: Highly classified US national security communications network goes down, sources say • CNN Politics

Natasha Bertrand, Katie Bo Lillis, Sean Lyngaas, Davis Winkie and Zachary Cohen:

»

The highly classified network that the National Security Agency uses for its internal communications is down, according to six people familiar with the matter.

The network, called NSANet, allows employees to communicate classified information internally. At this point there is no evidence of a cyberattack, sources said.

As the agency works to understand the scope of the problem, it has enacted staggered arrivals for personnel, the sources said.

“The National Security Agency experienced temporary degradation during scheduled routine maintenance,” the agency said in a statement on Tuesday afternoon. “While some internal systems were affected, mission-critical capabilities remain fully operational and our ability to defend national security continues. NSA’s infrastructure is designed with multiple layers of redundancy to maintain operations during planned maintenance or unexpected events.”

A source familiar with the matter told CNN that restoration of the network was ongoing but not yet complete.

«

Who’s next?
unique link to this extract


Alice Huang’s husband built a data-centre empire backed by the Chinese state • WSJ

Lingling Wei:

»

Alice Huang and her husband, Kang Wei, operated on two tracks pointed at the same goal of vaulting China’s artificial-intelligence capabilities ahead of the US.

Her job: cultivating a relationship—one that turned romantic—with Nvidia’s then-CEO Jensen Huang and securing a pipeline of the company’s advanced-computing chips that Washington has sought to keep from China. His job: building a sprawling network of government-backed data centers across China, a review of Chinese public records shows.

The scale of Kang’s data-centre business was striking. From 2021 to 2024, the proposed investments tied to Kang’s network topped $9bn, all built with government backing. Chinese government officials hailed Kang’s projects as part of China’s national push to ramp up computing power across the country.

Details about Kang’s work for state-backed AI efforts in China haven’t been previously published.

The couple’s work has spilled out into the open after news broke that Jensen Huang and Alice Huang were romantically involved and that the US suspected that she was providing information to Chinese intelligence agencies. Jensen Huang has since resigned from Nvidia as US authorities investigate whether export controls on Nvidia’s most powerful chips were violated.

At least some Nvidia chips obtained by Alice Huang’s company, Megaspeed, were shipped to China, The Wall Street Journal has reported, citing people familiar with the matter. Chinese public records, including securities filings and government announcements, don’t establish that Beijing directed Megaspeed’s procurement of Nvidia chips or that the chips ended up in data centres linked to Kang.

«

Sure, they don’t establish that, but you’d be mightily surprised if that wasn’t what happened. In fact, you’d think one of them had screwed up if not. Meanwhile, Nvidia recorded huge growth in revenue in Jensen Huang’s final quarter, said it didn’t want him back, and guided to strong growth in the near future. The latter remains to be seen.

In passing: “Details about Kang’s work for state-backed AI efforts in China haven’t been previously published” is a long-winded (American journalism) way of saying “EXCLUSIVE!”. (Gift link.)

unique link to this extract


A field of physics, audited by machine

Casey Handmer looked into publications in astrophysics:

»

158,059 papers and 554,670 citations. Every paper with full text screened, 659 of the most important recomputed and checked, and every verified error followed through the papers that cite it. This is the map, and what it says about whether science corrects itself.

I spent five years in this field doing my PhD at Caltech, and a few months of that went to errors in the literature that were already known. The equations my code was built on came from Bishop, Gómez, Lehner, Maharaj & Winicour, “High-powered gravitational news” (1997), the paper the subfield’s characteristic codes descend from. People knew it had mistakes. The fixes were scattered through some 10,000 lines of intersecting email threads that contradicted each other, and the paper itself was never corrected. The best I could do was correct the record in my own papers: an appendix setting out the full set of equations, made consistent, and a parenthetical “with a sign error in s3 corrected” for anyone reading closely.

That is the limit of human literature review. Mine was not deep: my three thesis papers cite fewer than a hundred papers between them, in a field of tens of thousands. Nobody re-derives every equation they cite, referees check one paper once, and nobody is paid to go back and fix the original. So every student who builds on a broken paper pays the same tax, one at a time, and the correction lives in somebody’s inbox. A machine can read the whole field, recompute it, and follow each error to every paper that copied it. So I had one do it. It found errors in my own papers too.

«

Handmer argues – and it’s hard to disagree – that scientific publications are over. AI means people can abuse the system by writing junk, but we can also see that peer review has failed repeatedly in the past.

The new age is aching to be born; we just aren’t quite sure what it will look like. If AI can review everything, why have peer review? Why publish some things rather than others? And especially, why charge insane prices for publications when anyone can publish them online?
unique link to this extract


‘Nothing they said made sense’: down the rabbit hole with Britain’s sovereign citizens • The Guardian

Mark Blacklock:

»

Wherever there are sovereign citizens, there is talk of Magna Carta. The charter, which was presented by 40 barons of England to King John in 1215, retains totemic status as a legal statement of the freedom and rights of the individual, despite the fact that 59 of its 63 clauses were either dropped during early revisions or have been repealed by UK acts of parliament. It was a specialist subject of Roger Elvick, who addressed an Aryan Nation meeting in 1984 on the topic, speaking for several hours. In Britain, sovereign citizens often prefer to call themselves freemen on the land, a term derived from the protections afforded to every “free man” in Magna Carta. Sîan Jones, a lawyer and senior civil servant at the Ministry of Justice, dates the freemen’s emergence in Britain to the late 2000s. “When we first encountered them,” she wrote in the magazine of the Justices’ Clerks’ Society, “it was initially a struggle to work out how to deal with them as nothing they said made sense, and engaging with them on our terms clearly wasn’t working.”

The earliest legal case I could find reported in the news was from 2010, when an odd-job man who had not paid council tax told Norwich magistrates court that he had joined the “freeman on the land” movement and revoked his consent to be governed by human laws. He was sentenced to three months in prison. Some of the early British freemen I found were associated with the Occupy London movement, the camp established at St Paul’s Cathedral in 2011 to protest against the global financial system. I corresponded with one who had run a popular website and given a talk at the Occupy encampment. He was not keen to talk to the press, and is no longer involved with the freeman on the land movement, but believed that the financial system was “on its last legs”.

In her book Doppelganger: A Trip into the Mirror World, Naomi Klein makes the case that conspiracy beliefs should be understood as holding a mirror to material conditions. The feeling of being failed by inhuman structures is all too real. We really are being let down by governments and institutions that fail to protect us from the worst excesses of capitalism. It’s simply that conspiracy systems imagine shadowy cabals in order to place the blame for complex social problems at the feet of particular groups. They give their followers illusory feelings of agency – that by doing their own research they can uncover what is really going on.

«

Utterly absorbing piece, with the scary detail that “more than a third of councils in the UK – at least 131 of 382 – have webpages warning residents against using sovereign citizen-style arguments to try to get out of paying council tax.” It’s the doltish mind virus.
unique link to this extract


iOS 27.3 beta adds daily Siri AI limits and iCloud upgrades • Mac Rumors

Juli Clover:

»

With the latest iOS 27.3, iPadOS 27.3, and macOS 27.3 betas, Apple is starting to test limits and paid upgrades for Siri AI and Apple Intelligence features.

Every Apple device gets a free daily access allowance for Siri AI and Apple Intelligence features that rely on server-side AI processing, such as Image Playground and Clean Up in Photos. When that limit is hit, there will be an option to extend usage with an iCloud plan. Here’s what will be limited after the daily allowance has been reached:

• Siri AI for multi-step, personal context requests that rely on server-side processing, such as asking Siri what time you need to leave from home to make it to a location on time
• Image generation and editing tools, including Image Playground and the AI options in Photos like Clean Up, Extend, and Spatial Reframing.
Workflows and automations in Shortcuts that rely on AFM Cloud and AFM Cloud Pro
• Third-party apps that use Apple Foundation Models in Private Cloud Compute for AI features.

While more complex, multi-step requests will be restricted, basic Siri commands will continue to be available even after a device limit is reached. Examples of basic Siri commands include those with a single ask, such as “set a timer for 30 minutes” or “text Eric I’m on my way.”

«

Jeez. For the first one, you could just open a phone and look it up on Apple/Google Maps. True, there are daily limits for the free models from other companies too (except Google?). This somehow doesn’t feel like the brave new world of Siri AI that we were half-promised though.
unique link to this extract


I leaked credentials on purpose for four months. Here’s who took them, and what they did next • THOR Collective Dispatch

Lauren Proehl:

»

TL;DR

• Stolen cloud keys are mostly used to send messages. 72% of the 155 AWS keys that attackers used went straight to SES/SNS: send quotas, verified senders, SMS settings. Exfiltration barely registers (S3: 5%)
• When a stolen credential looked like it could send, they tried to send. Attackers logged in to my fake mail server with the bait SMTP login and tried to push phishing campaigns through it. The fake server accepted every message and delivered none. The campaigns imitated Belfius bank (Belgium), a Greek “update your details” lure, and a Portuguese “account at risk” warning. There were also 419 advance-fee scams
• LLMjacking went from a side business to half the market. 12% of keys first used in August were checked for Bedrock. In September and October it was about half, spread across about 20 unrelated networks
• Median 37 minutes from theft to first use (n=103). A third arrive within 5 minutes. The slowest took 37 days
• The people who steal keys are rarely the people who use them. Only 12% of keys were first used from the IP that stole them. 82% were used from more than one IP; one key was tried from 1,691 addresses. 96% of the IPs that used a key never touched the honeypot at all
• Roughly 1 in 3 stolen keys gets used. That held steady across four months
• One stolen key, followed for 18 days: 75 IPs in more than 20 countries, from a quick “can it send?” check to a full inventory of the cloud account, and back two weeks later for the storage buckets
• Stolen AI keys get graded for resale. A checker tested five AI providers’ keys in five minutes, from 20 home connections in 14 countries, and slipped in questions designed to catch a fake endpoint like mine
• The exploit traffic was a botnet recruiting. 1,170 attempts from 161 IPs carried the same self-spreading installer. Others tried to install a cryptominer or steal the server’s own cloud credentials
• The attackers are cloud tenants; the people using the keys look like home users. 85% of scanning IPs were datacenter hosts. 74% of the IPs that used a stolen key were residential broadband, and 81% were in the US.

«

The existence of a trade in stolen credentials isn’t news, but it’s always helpful to have proper empirical research.
unique link to this extract


US loses measles-elimination status, erasing decades of progress • NBC News

Aria Bendix:

»

Measles is no longer eliminated in the United States, reversing one of the country’s greatest public health achievements this century.

An independent commission of the Pan American Health Organization (PAHO), a group within the World Health Organization, revoked the US’s measles-elimination status, according to sources familiar with the vote, after determining that the virus has circulated nonstop for more than a year.

The US earned the designation in 2000. Now, measles is once again endemic.

PAHO is expected to officially announce its decision on Tuesday.

“The United States has an extraordinary measles problem. We’ve had more measles cases in the last 18 months than we’ve had in the previous 25 years combined,” said Noel Brewer, a public health professor at the University of North Carolina at Chapel Hill.

Upwards of 5,400 cases have been recorded this year — more than double last year’s count and the highest since 1991. Ten people have died, according to state health departments.

«

Great job everyone, well done.
unique link to this extract


• Why do social networks drive us a little mad?
• Why does angry content seem to dominate what we see?
• How much of a role do algorithms play in affecting what we see and do online?
• What can we do about it?
• Did Facebook have any inkling of what was coming in Myanmar in 2016?

Read Social Warming, my latest book, and find answers – and more.


Errata, corrigenda and ai no corrida: none notified

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.