676 points | 5d ago | Discuss on Hacker News | Back to Radar
Currently says "Partial Outage Affecting Recommendations Tracking and Reporting"
> Your origin can be clean while your visitors are robbed — the compromise sits in the tag layer.
> On mobile, Report URI says they attempted to install exploit kids.
The models are using language correctly, but their whole universe is limited to a single session. They're not aware that all the previous sessions also used the same phrase, that you've seen it 10 times in the last hour, etc.
Humans naturally avoid repeating the exact same phrases and jokes they just heard from several other sources, but sometimes you can see the same phenomenon with people too: when there's new bullshit phrase or trend spreading through tiktok/instagram (or popular press), there's a brief moment you can catch many normies regurgitating the exact same phrase "independently".
The front of the meme virality wave sure does look weird.
It’s not high-quality, though — it’s SEO speak to pad blog posts without adding anything of substance. We notice it more because that stuff is so common in the training data but not the experience of anyone who doesn’t spend time on content mills.
Anyone knows what this might be? Can you jailbreak from safari on those versions?
[edit] From the other link, it seems to be Coruna and DarkSword exploits
The official advise is to keep a check on your account history some time before November … lol.
Can't wait until they start killing hardware
https://x.com/tehlike/status/1432082161125888003 https://x.com/tehlike/status/1467033965215109122
Somehow this sentence immediately flags this article as AI slop.
What is interesting is that the write access wasn't just for a single customer, but hundreds. So either tenancy isn't strong enough, or the level of access involved in the compromise was pretty severe.
whereas the one that we are commenting on now looks like an ad: https://blog.report-uri.com/worlds-largest-brands-infected-w...
It opens with:
> This is exactly the kind of attack Report URI is designed to detect. Our platform can ...
It's otherwise nearly the same information.
It would be kind of ironic if the hijacking warning article got hijacked, I guess?
2. The former seems super more clear to me to be AI generated prose.
3. The latter explicitly continue talks about the actions they took, that they reached out to the affected parties, etc. From a quick scan, it does not appear that the former claims to have done anything themself?
I hope folks are re-evaluating that now.
That’s why it needs to become criminal for a website to serve malware, including through third party channels like ad networks and analytics services. Serving shit websites is no different than knowingly, or negligently, dumping toxic waste and should be treated the same.
We should also enforce antitrust laws.
Open source is already expensive from a time standpoint, you wanna make it expensive from a monetary standpoint too?
There should be standards that people have to live up to if they are going to provide a product, even for free, to others.
If your website is making any requests outside of the original domain, there's no "accidental" in anything the website does.
I see no way to distinguish between huge megacorps and individuals.
I never really understood this idiom, because how can a PHB engineering manager (for example) be required to not understand their team's technology in order to retain their job? Surely if that manager knows their team's work inside out then they would be a more effective manager - which is good for everyone?
If I squint-my-brain, I can see how (for example) that anyone who works full-time as an SEO "consultant" in the year 2026 would only be if they actually believed that SEO tricks actually worked, which requires them to not understand how modern search-engines' spiders actually work (because it's my belief that "SEO" has been either nonsense bollocks - or a complete grift - since the mid-2010s). Am I close?
In this case, it's the dangers of the tracking and advertising frameworks being used for hostile purposes.
It's a continuous case of whack-a-mole. SEO consultants think <thing> would be a good way to improve the search ranking of their customers, it works, it spreads throughout the industry, Google catches on, rinse and repeat.
As soon as something is a KPI, there is incentive to game the system - with the top end of "gaming the system" toeing or crossing the line into fraudulent, anti-competitive or otherwise illegal territory.
I know we really like mass incarceration in America. But maybe start with civil liability and a private right to enforcement?
--Upton Sinclair
The follow on line from that one is even better: "The most deluded people are those who choose to ignore what they already know". I would probably want to add on things like "because it's not convenient" or similar for particular relevance to some of the stuff I've been dealing with recently at work. Folks know it's dangerous to do something a certain way, and know how to do it a safe way, but it's not as convenient so they delude themselves it'll be fine just this once. And just this once too. And just this once, once more. Slowly but surely normalising the deviance.
In the context of the Optimizely case, people know they're giving it dangerous levels of capabilities, but it's just too darn inconvenient to do it in a safe way so sure, why not. It'll be fine (and it will be, right up until it isn't and you realise your choice has compromised your users)
We got 10MB of javascript (maybe 10 years ago...) because engineering needs to build one frontend with multi browser and multi screen support so they picked an off-the-shelf do-everything framework like Amber or Angular, or a react/redux build with 300 nested dependencies.
The most infuriating thing, when you do remove these tags, marketing forgets they exist and are like "Oh we don't use that anymore".
Often also true for software development, unfortunately.
I can imagine the big corporations spending millions on econometrics teams or commissioning mix modelling like the olden days, but it's pretty shit by comparison, and beyond the MNCs there's hundreds of thousands of SMEs without the money/scale and with business models that hinge on digital advertising.
Life's too short to work for people who don't care about others.
and things continue being insecure
Life’s too short to care about random strangers on the internet sometimes.
Then you've done your job and move on. Just make sure that there's enough public record of your warning that - when the next cdn supply chain attack happens - you can say "I warned you about this and you decided not to take action".
No organization is perfect. I don't think this is something to quit over. Just make sure to CYA.
A javascript file served from example-a.com should be able to run on example-b.org only if it has been signed with the same TLS key used for example-a.com .
NPM isn't perfect but there's typically a substantial delay between the release of a malicious package and that code getting onto your production environment. Unless you're automatically updating every day, there's a good chance someone will discover the problem before it affects you. And you can always scan your upstream dependencies.
This doesn't work if you're using someone else's CDN. Any change to their content is distributed to all your users immediately.
https://developers.google.com/tag-platform/security/guides/c...
In the end this is a tradeoff between security and convenience.
Oh how I hate that one. I wish Facebook, Instagram, Google and others would figure out serving the public scripts from .com only.
Not going to happen. Most decisions in tech do not factor in risk, only cost. Risk mitigation usually requires considerable spend, and very few people/companies are willing to spend on low-probability events.
AI-assisted hacks and outages are accelerating the process of making the costs of weak security apparent.
I'm programming for over 30 years now and GTM is my personal #1 most hated software.
They knew they could do it, and how, they just didn't want to spend 10 minutes doing it themselves. To them, asking an engineer to do it in code is easier because it only takes a minute to ask.
Even one of these accounts being compromised would be a massive security incident that would potentially put millions of people's data at risk.
In my experience the vulnerabilities are coming from tools brought in by less technical teams, which then get misused by engineers.
These tools are oddly "safe" in the hands of Marketing and the like, because nobody trusts Marketing, so they stick a bunch of guardrails in place to stop anything stupid from doing serious damage.
The same tools are dangerous in the hands of engineers - many engineers trust themselves to do risky things, and will bludgeon you with debate if you tell them they have to follow the same rules as those _simpletons_ in Marketing.
They're usually client-side because engineers don't want to waste time implementing things for Marketing (or more accurately, struggle to support a domain that is radically multidisciplinary and just-in-time and revolves around conforming to non-engineering industry norms, to such a degree that everyone prefers siloing themselves), and the safe space given to Marketing is usually a client-side tag manager so that's where they do it when they have to do it themselves, and then once the tool is implemented client-side, engineers piggybacking off it aren't going to bother with doing a new server-side implementation.
Cue engineers loading Optimizely in their secure product areas, getting exceptions added to CSPs, and the like. Then a major incident. Then a post-mortem highlighting the risks of letting Marketing manage SaaS, then a project to make a secure in-house experimentation solution which the engineers will abandon shortly afterwards... and then finally, a couple years later, someone will discover Marketing has procured Optimizely 2.0 and the cycle repeats.
Unless we get rid of product managers, which isn’t a terrible idea.
When IDs don't persist across domains or subdomains there are also plenty of methods that don't involve passing via URL or tripping reasonable network security, and that work for >99% of user clients.
We get sanctimonious about this, but geeks build all that monetization infrastructure, not people from marketing.
First thing I add to my agents.md is do not use third party libraries, frameworks, fonts, icons, etc unless strictly necessary, ask first
In this case optimizely should have had better systems to spot a release like this (internal cool downs, check and balances, phased rolouts, automatic security checks). And sites that have massive user bases (cnn, bbc, etc..) have zero excuse to not have the same.
I would go so far as to say those sites should not be serving from vendors but should be vendoring their own scripts that they also are passing through security audits and serving them selves with cool downs.
How do these companies not have better dedication to protecting their users is beyond me.
It doesn't cost them anything but a "whoops" post when they fuck up, so the incentive is misaligned (as you noted under repercussions).
The cost is externalised onto everyone else not them so it's not going to change.
Who cares? Next week you forget about this and move on
This is the bare minimum. Automatic 3rd party requests should be considered data unsolicited data disclosure under GDPR.
A company making say bricks has an incentive to cut costs too certainly but they’re not implementing loot boxes, tracking people semi covertly, implementing dark patterns to boost engagement by any means or heaven forbid launching missies based on sketchy AI decisions.
The incentives are fucked across the board. When you look at the CEOs we revere and the ones who have the most leverage and mobility, it’s shocking to see how many absolutely destroyed their company in exchange for a good few quarters. That’s actually seen as good somehow, those decision makers are desirable!
When Europe tries they are either laughed at or accused of slowing down innovation.
The code should be static and the configuration (data-only) the only thing that changes frequently, but that is less flexible and harder to implement.
I would go so far as to say those sites should not be serving from vendors but should be vendoring their own scripts that they also are passing through security audits and serving them selves with cool downs.
How do these companies not have better dedication to protecting their users is beyond me.
It's not all or nothing. NoScript [0] lets you run JS (and other specific types of content) from domains and/or sub-domains you allow and reject from those you don't. So you'd allow JS from my-serious-business domain and their CDN, and disallow all else, including those mentioned in the linked article.
It would not have helped here. gitlab.com and my bank are on the list. Obviously I could not have blocked them.
Per the linked article, those are not the domains where the malicious JS was served from.
Was a pain though on eCommerce sites as they'd often switch payment processors and then you had to do the whole dance of partially successful payments until you found the right set of allow lists.
in some cases they were
To be more specific
NoScript would have still caught if such proxy was through
* a separate custom domain that was not previously trusted by the user
* a separate sub-domain of a trusted peer domain (like the opt.scheels.com example) and NoScript user opted for sub-domain level blocking
NoScript would not have caught if the proxy was through the first party primary domain (like the www.vodafone.com.au/js example) - i.e. the domain or sub-domain that the user had already trusted.
So, while it's not a full guarantee, it reduces the exposure risk significantly in most common scenarios.
“A full inline copy of the payload, which runs through eval if 1688360.cc is blocked.”
No, the article doesn't state that the JS served by all listed first party websites was compromised (but see note below). The original maliciously modified content was still (in most cases) served from an optimizely domain like cdn.optimizely.com. The cdn.optimizely.com would not have been blocked by web browsers. The "full inline copy of the payload" refers to content served through that optimizely payload, instead of those additional domains the attacker had set up. About a paragraph or so above your quote:
> The file was still served from cdn.optimizely.com, with Optimizely's certificate, from a host that every affected site already trusted.
That said, while this is a very common scenario, it doesn't cover all of them. There's still a possibility that certain optimizely customers have set up a reverse proxy for the optimizely JS snippet. If this proxy was through a separate sub-domain it's possible NoScript would have caught it (if user opted for sub-domain level blocking). If this proxy was through one of their own primary domain names that user already trusted in Noscript, it would not catch it. So it's not perfect, but covers most common scenarios.
Taking a step back, I'm not sure why there is so much pushback in this thread. Most of it seems to be based on either no or incorrect reading of the information or incorrect understanding. I'm not in any way associated with NoScript or trying to convince folks to use it. Given the incident, I think it's worth stating that this type of attack falls under its main use cases. A reasonably configured NoScript would have protected users in most of these cases. Whether it makes sense for users to use it or not is up to them - it depends on where you see the line between security and convenience tradeoff.
I don't doubt that. But how many users can achieve a reasonably configured NoScript, without breaking things they want to use? I am a developer, so I should be able to do it or I were a bad one. But I am sick of an immoral industry that requires me to do extra work because they are trying to steal something from me in an ever growing multitude of ways every time I click somewhere and even without.
Allows for really easy, selective and contextual permissions for some critical capabilities.
Interesting that it is not necessarily a supply chain attack a la LiteLLM/some of the other recent breaches. Possibly just a mass compromise of customer passwords.
I guess all I can do now is emptying my browser cache to be on the safe side. The writeup (piece of marketing for their product) has no section for affected end users
Yes, I sometimes used NoScript and uBlock Matrix, but gave up on those because it's a tedious job to manage your allow lists.
0.0.0.0 cdn.optimizely.com
...and now I have another reason to keep it there.From what I could tell, the malware just generally looked at anything on the page, in cookies/local storage, or clipboard that looks like a crypto address or credit card number. So for sites like CNN or BBC they shouldn't be able to do much damage unless there is a subscription page for example that takes credit card number.
However, the malware also delivered the Coruna and DarkSword iOS exploit kits, which do take advantage of browser vulnerabilities in iOS to cause more serious damage.
The model is broken on fundamental level and it will never be fixed.
I think it's not used because they are continuously updating their scripts for users.
An SDK doesn't necessarily mean that scripts are kept up to date as a site's build could've been done months or years ago with the SDK version pinned in npm.
If I were to setup a system today, I would mirror any script from a third-party and pipe it through a LLM to check for malicious code before deploying.
Optimizely, or Episerver as it was previously called, was such a pleasure to work on - but since version 13 they more or less vendor locked you into their own authentication (Opti ID / Entra AD (?)) and their Optimizely Graph for search and content look up. You can keep self-hosting but you won't get access to Opti Id or Graph if you're not on their tenant.
I guess this is the inevitable happen - everyone gets hacked at the same time when hosting on a SaaS platform.
All of that for them to mess their own security.
If it was on me, I would stop using them.
Isn't this anomalous? using RCE root chains in mass-delivered malware? Seems like a new precedent.
Citizen Lab’s website lets you browse trough all the malware they’ve found, so search and ye shall find lol.
So is this article just an ad? It makes it hard to tell what's true and what's exaggerated.
But how do you know that you personally didn't get affected? Check here: https://karlnorling.github.io/optimizely-hack-checker/
It’s the same reason financial analysts build the tower of bable in Excel. Some COULD access engineering resources, but they don’t want to. The fact that there’s no controls, audibility, or quality management is not a “bug”, that’s the primary feature they’re after.
This also describes the risk of giving non-engineers access to runtimes and AI. They can, and will, build monstrosities outside of the purview of management.
Apparently, one of Optimizely's core features is its WYSIWYG editor. This means it's supposedly useful for running A/B tests without writing any code.
However, I don't know what kind of code is generated by WYSIWYG, and I don't have a good impression of it.
And would uBlock origin have prevented it?
Some ublock lists (e.g. Peter Lowe's server list) block everything from optimizely.com, so if you enabled those you're good.
For anyone unaware, this also works on Firefox on Android, as it has support for almost all extensions.
https://www.newsweek.com/cyber-attack-websites-affected-1272...
Unapologetically AI/LLM. I reckon that’s where I and 80% of readers stop being readers.
Search for BBC Search for CNN
Both were part of the research and were audited as recently as October 2026 with evidence recorded that they werre deploying Optimizely without consent (a legal obligation in the EU).
So smoking gun evidence that they were unlawfully deploying Optimizely in the last 6 weeks across the EU leaves them ripe for litigation.
Comments are loaded live from Hacker News and are not stored by Mid or Real.
duxup 5d ago on HN
“yeah load this JS m, but it can’t load / random other JS referencing sketchy places”
1a527dd5 5d ago on HN
duxup 5d ago on HN
stavros 5d ago on HN
technion 5d ago on HN
Common examples like analytics tools happily change their script bundles whenever they feel like it without just tagging a new version and letting you upgrade.
I wish SRI could be used more,
skeeter2020 5d ago on HN
computerfriend 5d ago on HN
mook 5d ago on HN
nchilada 3d ago on HN
SRI (or CSP hash, or vendoring) only helps if someone actually reviews the changes instead of automatically approving them. Maybe it's more likely to happen with LLM reviewers, though in such a world, one can also just ask said LLMs to help manage in-house source, or to help leverage a properly versioned and auditable library, to test all the different shades of slop.
Any website that cares about security should have long since moved on from unpinned JavaScript controlled by a third party. It may have made sense 15 years ago, but certainly not in 2026.
(disclaimer: former Optimizely employee)
skeeter2020 5d ago on HN
cowlevel 4d ago on HN
r1ch 5d ago on HN
Jach 5d ago on HN
Feels nice knowing several years ago I worked on forcing CSP (including removing unsafe-eval) on some customers for a site builder product, as they have to manually whitelist domains which limits the impact of these sorts of attacks...
This writeup https://blog.report-uri.com/worlds-largest-brands-infected-w... has a section on how CSP helped (or didn't) in this attack.
horsawlarway 5d ago on HN
As an aside - I'm rabidly against "analytics" companies that come in and suggest that the right "getting started" choice is to drop untrusted JS they own (and can update at any time) onto your site.
It's "easy", but it completely breaks your security model.
If I were a lawyer... I'd be taking a serious look at putting optimizely into the meat grinder for recommending 'unsafe-eval' as a script source... it literally has "unsafe" in the fucking name for a reason. This alone should end the company. It won't, because this space is absolutely FILLED with scummy behavior, but it should.
---
A strict CSP is one of the most meaningful ways you have right now as a web developer to reduce attack surface. People should take advantage of it.
There's a reason browser vendors now force basically all extensions into a strict CSP by default... https://developer.chrome.com/docs/extensions/reference/manif... They won't even let you manually relax it anymore.
cowlevel 5d ago on HN
horsawlarway 4d ago on HN
So are equivalent companies that recommend similar strategies. Sprig, segment, amplitude, launchdarkly, heap, full story, etc...
It's a very long list.
They all operate on the same strategy. Pitch to execs that they solve real problems and they have basically no eng cost to implement.
All you need to do is inject them into your page with absolutely no restrictions and let them run any scripts they'd like.
cowlevel 4d ago on HN
nchilada 3d ago on HN
Softer questions might be: why is eval (still) being used for any inputs at all; and to the extent that it is used, why did¹ the docs not concede the breadth of CSP unsafe-eval (which allows eval in all scripts, not just Optimizely's), and why do the docs actively discourage attempts to enable Trusted Types (which includes the much narrower CSP trusted-types-eval²) instead of actively facilitating its use?
And needless to say, a more direct question would be: why should customers use a class of product that is so powerful that unsafe-eval doesn't increase its capabilities at all?
¹ cf. https://web.archive.org/web/20260810141143/https://support.o... vs. https://docs.optimizely.com/web-experimentation/docs/update-... (updated yesterday).
² ~84% browser support, per https://caniuse.com/mdn-http_headers_content-security-policy...
nchilada 3d ago on HN
CSP script-src was only helpful here because the attacker used the Optimizely ProjectJS feature to delegate to a third-party script. It wouldn't do a thing if the dirty work had been done in e.g. ProjectJS itself.
CSP connect-src/form-action are somewhat helpful for stopping a script from exfiltrating data to, or redirecting the end-user to, a third-party domain, but an attacker still has plenty of other options (https://github.com/w3c/webappsec/issues/656).
And apparently CSP frame-src would have blocked the iOS exploit, but there's nothing to stop other malicious content (scam propositions, pump-and-dumps, psy-ops, other “truths”...) from being rendered.
(disclaimer: former Optimizely employee)
JoshTriplett 5d ago on HN
Never load third-party JavaScript on your site. Zero exceptions.