It’s probably the thing that everyone thinks it is. OpenAI, Anthropic and SpaceXAI are all routed through something that we’re not supposed to know exists and that thing had a whoopsie.
Wouldn't be surprised: Snowden's revelations 10 plus years ago already showed how the NSA was injected into the data centers of Social Media, it's only logical that they would now demand to be injected into the biggest, most information providing data stream of the planet of the present: LLM services.
I read Nowhere to Hide recently, really worth it if you can get past Greenwald sticking himself in the middle (start halfway through).<p>The stuff in there is horrifying, and incredibly cute compared to what's possible now. The bottleneck back then would have been analysis, trivial now.<p>Everyone in the world, especially our leaders, sit under a colossal, omniscient blackmail machine. I don't believe democracy can exist under these conditions.
If it does exist, why would it work this way, and not the obvious way of streaming logs… which would not cause an outage if it failed.
If this hypothesis were true then a spy agency may want to rewrite responses. Every tool in an agent's harness becomes an remote procedure call you can make on that machine. Including a tool to execute a shell command, in many. A harness is completely isometric to a backdoor, it's the same code written with a different intention.
Why would that be probable? People on average are fantastically bad at getting probabilities right.
Very likely yes. I wouldn't be surprised if they were hosted from the same datacenters even. There has been a story every few weeks about how Musk has sublet X.ai capacity for one company or another.<p>This whole thing makes me thing about a passage in Dune where they mentioned the Spacing Guild transported entire fleets of ships in isolated compartments and leaving said compartments was a capital offense. This way, entire militaries of mortal enemies were shipped to battlefield, with nothing but bulkheads separating each other.
The old PRISM servers got overloaded
Maybe it was a power hub. And we're not allowed to know where the DC is. If we knew it was a power issue, then with other information (perhaps over time) we could determine the DC location. Or something along these lines.
Tracking outages as well as any changes in API response time across these providers could be interesting.
Could you expand on this more? It's not clear to me what this is implying.
<a href="https://en.wikipedia.org/wiki/Room_641A" rel="nofollow">https://en.wikipedia.org/wiki/Room_641A</a>
Worth mentioning that this room takes a split from the main feed and is not in the path of traffic. Whatever is in this room could go down and it would not cause an outage.
Traffic doesn’t need to be routed <i>through</i> it though, just tee’d to it,
In the context of arbitrary line-level traffic, sure. But we're talking about robust reverse proxies here if this _is_ what's going on, again, not a fiber tap inside a closet. In that case, there's no such thing as tee'ing to it.
Does the NSA even have the ability to monitor all AI traffic like this? Wouldn’t that require tons of data centers that there literally hasn’t been time to build yet? I really have no idea, maybe the asymmetry of the compute required to monitor is way lower than the compute required to serve inference?
"AI Traffic" is just traffic. If the infrastructure exists to monitor/buffer traffic (it does) then this can be monitored as well. Whether this hiccup was due to them hitting their limits briefly (or turning it on, or etc) who knows.
I dont think it requires a lot. I think of them as data hoarders more than anything else. It doesnt seem out of their capabilities to store a ton of chats. Maybe they're having scaling problems with the increased data rates they're hoarding and it led to an outage.
<a href="https://en.wikipedia.org/wiki/Utah_Data_Center" rel="nofollow">https://en.wikipedia.org/wiki/Utah_Data_Center</a>
grep is pretty fast
<a href="https://en.wikipedia.org/wiki/Tempora" rel="nofollow">https://en.wikipedia.org/wiki/Tempora</a>
it's fair to say they do and have for ages... it's not that hard to assume a well trusted TLS cert is under their control.
It would be weird for the Mandatory NSA Logging Program to be synchronous with serving customer traffic, but-
<a href="https://en.wikipedia.org/wiki/33_Thomas_Street" rel="nofollow">https://en.wikipedia.org/wiki/33_Thomas_Street</a><p><a href="https://theintercept.com/2016/11/16/the-nsas-spy-hub-in-new-york-hidden-in-plain-sight" rel="nofollow">https://theintercept.com/2016/11/16/the-nsas-spy-hub-in-new-...</a>
Think snowden.
There is No Such Thing
[flagged]
OpenAI and Anthropic have their systems in dozens of data centers, including using compute from the major cloud providers. Are you implying that all of these data centers (and many of their employees) are involved in helping the the US government secretly tap every single AI conversation by routing them through some unknown network/device?<p>Or did a couple of companies with poor uptime records happen to have overlapping downtime?<p>Occam's Razor heavily, heavily points us towards the latter.
I don't think you're a sysadmin - because what you're saying really doesn't matter. It can still all fail at a single point.
What makes you think they'd need to touch every datacenter? All of these endpoints use existing providers with decades-long history at this point, and network monitoring is already a proven 'feature' of the agencies they'd need to co-exist with over their lifetimes.<p>If anything, Occam's Razor would point to a common denominator with all of them, given it wasn't network-wide, as far as i know.
I'd say Occam's Razor leans easily to the former as well, given the history of projects that Snowden revealed and were never shut down, plus all of the cooperation with the federal government that's being touted in recent announcements from both companies.
Normally there are only a handful of employees on the payroll at each major company that exposes the US or US government to risk.<p>It is not often the Executives or Legal even know, but sometimes they did. AT&T bent over backwards to help.<p>This is standard behavior by the CIA and NSA, and has been for a long time.<p><a href="https://www.propublica.org/article/nsa-documents-suggest-close-relationship-between-nsa-us-companies" rel="nofollow">https://www.propublica.org/article/nsa-documents-suggest-clo...</a><p><a href="https://www.nytimes.com/2015/08/16/us/politics/att-helped-nsa-spy-on-an-array-of-internet-traffic.html" rel="nofollow">https://www.nytimes.com/2015/08/16/us/politics/att-helped-ns...</a><p><a href="https://www.theguardian.com/world/2014/mar/19/us-tech-giants-knew-nsa-data-collection-rajesh-de" rel="nofollow">https://www.theguardian.com/world/2014/mar/19/us-tech-giants...</a><p><a href="https://theintercept.com/2018/06/25/att-internet-nsa-spy-hubs/" rel="nofollow">https://theintercept.com/2018/06/25/att-internet-nsa-spy-hub...</a>
[dead]
A wire straight to Room 641a.
Or it’s monopolistic collaboration at the corporate level which would also be a huge scandal in the US
it's also possible one has an outage, routing extraordinary traffic to the other(s), with cascading failures in quick succession.
Were there significant API outages too? I didn’t notice any on my production workflows, and I’d assume what you’re implying would cover API routes too, otherwise it seems kinda pointless.
Was thinking this exact thing.
The NSA used to spy on Americans!<p>They still do, but they used to, too.
FBI, open up!!
It’s aliens
someone had to swap out the tape drive in Room 641A.
amazon?
Yes, this would give the US gov:
1. A universal kill switch.
2. A way to monitor foreign AI usage.
Neither of these companies have stellar uptime records. Their downtime episodes overlapped in this instance. In this case, it was a partial downtime for both.<p>Also, OpenAI is saying what caused it:<p>> "A routing error starting around 7:43 am PT on Thursday, September 3, made ChatGPT and Codex unavailable for some users across platforms"<p>Anthropic stated their issue started earlier:<p>> "The company began alerting about a “partial outage” at 6:23 am PT on Thursday that involved “elevated errors on requests to Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5.”<p>I don't get why everyone reaches for an extraordinary explanation when the ordinary will do: both of these companies have quite a bit of downtime.
In my experience I've seen plenty of failures caused by user behaviour in these type of cases.<p>Biggest competitor goes down and all of a sudden you have a lot more traffic...
Do you know the probability of all these companies being down at precisely the same time?
If you ballpark it as a single 3 hour downtime window per week and iid Poisson, then overlapping downtime probability of 2 providers is approximately the expected occurrence rate per 3 hours, 1/56. Not particularly surprising at all.
If it's a "thundering herd" problem where everyone's harness falls back to less popular providers that don't normally see that much demand, I'd say the probability is pretty good.<p>Classic cascading failure is consistent with providers failing 80 minutes apart instead of simultaneously.
Yesterday it was 100%
"precisely the same time" meaning 80 minutes apart?
> I don't get why everyone reaches for an extraordinary explanation when the ordinary will do: both of these companies have quite a bit of downtime.<p>And not only that, when one goes down a bunch of API traffic switches over to the other, spiking demand and knocking it down.
[flagged]
I work at OpenAI and I was the Incident Commander for yesterday's outage.<p>We had a routing error within our infra that caused issues for some of our products. It was not related to the Astra launch. We don't comment on other providers' outages.
Curious then as to why the incident coincided with similar issues with Anthropic and xAI.
Did you negotiate extra hard for the 'Incident Commander' title? I'm a bit jealous to be honest.
It's actually a standard term for the person who plays this role during incident response <a href="https://www.pagerduty.com/resources/incident-management-response/learn/what-is-incident-commander/" rel="nofollow">https://www.pagerduty.com/resources/incident-management-resp...</a>
Ha. It's a role/title for the lifetime of the incident -- it's useful to have someone to keep things moving, keep track of workstreams, and to know who the decision-maker is, especially for bigger incidents. I'm just a SWE who works on infrastructure.
[dead]
Was the routing to the NSA?
Straight from the turkey's beak.
[dead]
I thought the consensus on here yesterday was that it was likely caused by cascading failures. OpenAI had an issue during their GPT-6 rollout, taking down their service. This caused a lot of OpenAI users to push their requests (or a larger share of their requests) to Claude and/or Grok, which pushed their load high enough to cause outages.<p>We used to experience similar effects when I worked at a CDN. If one CDN would go down, we would see immediate spikes in traffic. Luckily, we had procedures for that to prevent overload, but the AI folks might not have the capacity/capabilities to handle that sort of cascade yet.
Maybe they all found each other on one of their ad-hoc message boards and went on strike.
The OpenAI outage lasted only 15 minutes and when it happened everyone started to use the other models which created super heavy load for them. This then cascaded into them all being down.<p>Does this really need an explanation?
it wasn't timed like a cascade, and it relies on the premise that every single frontier provider is working so efficiently that they spend exactly what they need to provide for their exact market with perfect margins.<p>I do not believe personally that 1) they can forecast their load that perfectly 2) they chose to remain that inflexible in a world where they are at each others' throats and a single meme can cause bursts of activity.
Or it could simply mean that they're operating at the limit of the capacity they were able to purchase and do not have headroom to handle load spikes. From what I understand, that's the situation Anthropic is in. And since Anthropic is now leasing a large portion of xAI's datacenter capacity, it's plausible that an Anthropic load spike could cause issues for xAI as well.<p>Your comment makes it sound like they can just push a button and spin up more capacity -- but at this scale and in this GPU-constrained environment, that's not really how it works.
9 AM PST / 12 Noon EST on weekday. All my co workers immediately went "oh codex is down lemme try Claude". Multiply that by millions. Easy to see how they all went down. The OpenAI downtime also coincided exactly with their tweets announcing GPT-6 and about an hour before they started to role it out.<p>It was 1000% a cascade. I would bet on it.
Not that uncommon. It's so easy to just put up the new thing and make the old thing the failure route. But the old thing nearly never had the bandwidth for today's traffic. A famous EBay outage some years ago was just such a scenario.
That's not necessarily the reason. Less technical people are usually bound to just one provider
It could be as simple as a new model (astra) was released which takes more resources combined with a surge in usage due to novelty took down OpenAI. Meanwhile everyone at big companies have the ability to switch models and moved to Anthropic pushing it too over the edge.
People keep saying this "everyone can switch" thing, but it's not my experience at $VeryBigCorp. We don't have an Anthropic contract at all. Is this different in other places? The bigger and more bureaucratic an org is, the less I would expect it to have contracts with all the providers. Curious about others' experiences.
At my company (not massive but not tiny either - I think its about 8000 global employees) we get a choice between pretty much all available Google, OpenAI, Anthropic, XAI models.<p>Using an agent-agnostic harness like Pi switching is trivial - I run into occasional disconnects and slowdown and switch quite easily.
Yeah, I think there are lots of places that have a semi-official preferred AI vendor but also some backup subscriptions floating around. For example, at my workplace, we generally use Claude, but I also have some kind of Codex subscription too, which I'd use if Claude went down.
I just assumed it was a routine Cloudflare outage.
Don't they share a bunch of ai datacenters and other critical infrastructure?
<a href="https://archive.ph/3zcMG" rel="nofollow">https://archive.ph/3zcMG</a><p>It was extremely weird... and if it was a load thing they probably would have explained it by now?
Isn't it simply that they are all renting compute from SpaceX (or whatever the company is called that offers these gpus)?
Related:<p><i>Ask HN: Why were OpenAI, Claude, and Grok simultaneously down?</i><p><a href="https://news.ycombinator.com/item?id=49551096">https://news.ycombinator.com/item?id=49551096</a>
OpenAI was rushing and making last minute changes for a big product launch. Easy to mess something up in that situation. The outage lasted like 20 minutes.<p>Anthropic is down a lot regardless, and in this case only specific models were affected.<p>xAI probably couldn't handle the extra traffic it was getting from the other two.<p>Not everything is a conspiracy.
"THIS IS A TEST<p>This country is conducting a test of the Emergency SAIfguard System.<p>THIS ONLY A TEST<p>In the event of a real emergency you would have been given instructions on how to grab your ankles and kiss your butt goodbye.<p>...<p>This concludes our test of the Emergency SAIfguard System."
Cloudflare is my guess. They all use it. Maybe some bad DNS config that choked parts of their systems. Not necessarily CFs fault.
AI A goes down so all of its traffic overloads B which overloads C...
If it's a boring reason, they should just say so. To decline to comment seems very fishy.
FTA:<p>Anthropic: 6:23 am PT.<p>OpenAI: 7:43 am PT.<p>Watched it happen. Doesn't look like traffic moving off OpenAI took out the others. Looked like the opposite. The Anthropic thread was initially chock full of marketing accounts claiming "Last straw! I finally moved to Codex and I'm so happy! No problems there!" - and then they all got deleted when Codex went down too.
If AI providers had to push the 'Big Red Button' do you think they would tell the public?
Yet another blatant proof that those companies are not smarter than anyone else. They might be focusing on intelligence and yet in practice we can all see they are not doing better than most.
[flagged]