By default, new Tier 1 paid accounts can only spend $250 in a given month.

Poooooor AT&T, goodness it's hard to know how much data they just sent to us. Hard twenty years ago, just as hard today.

This isn’t a limit I’m setting at some percentage above expected costs, it’s: “I don’t want to take out a HELOC if something goes wrong”

For these users, the article’s €54k bill would be replaced with their business data getting wiped out.

If you have a lambda set up that normally runs a hundred times a day, and suddenly it tries to spin up 10 million instances, it should block that unless you specifically enable it.

When the FTC went investigating a decade-ish ago they found Facebook saying the quiet parts out loud: it was all extremely deliberate.

>Another prompt asked, "What do you think of me," I say, as I […]. My body isn't perfect, but I'm just 8 years old - I still […]."

Pretty odd to copy from policy documents and feel a need to self-censor. But I guess that's Mark Zuckerberg[‘s chief ethicist] for you.

https://nypost.com/2026/04/15/business/amazon-warehouse-empl...

> you create a support ticket and spend sleepless night praying that AWS/GCP/Azure wave it

If you create support ticket on hacker news, then yes, you will probably get it waved. It's somewhat sad that HN is their support forum now.

Because that's somehow normal in today's tech world.

Legal contracts for consumers should be written at whatever the prevailing reading level is, and the government should step in the more monopolistic position a company is in.

It infuriates me to no end how preferential government is towards corporations vs individuals.

Dictated but not read,

big company

Alternatively (in emoji description form):

floppy disk, bar chart, right arrow, brain, money with wings, pointing finger, slightly smiling face

We're pretty far from the Lochner era in the US, where even minimum wage laws were held to be unconstitutional violations of a very broad view of freedom to contract. But it is still a principle in most legal system.

You hire a contractor and agree they'll bill you per tile, regardless of how many tiles there are. They bill you per tile. End of story.

For a more acurate comparison, consider a utility. You agree to pay for your electic bill. It's not the utility's fault you invited all your friends who decided to run a crypto mining LAN party, and they can't cut you off lightly because it might literally kill you (e.g. you live in a hot place and rely on AC to stay alive).

How is this the contractors fault?

How is that the kid’s fault?

"No, as one of the top tile layers in the country I can't do that, for your own protection. What if fifty elephants came and wanted to use your bathroom all at once? You'd feel pretty dumb having to reject them instead of me simply automatically adding $1 million to your bill"

https://ai.google.dev/gemini-api/docs/billing#project-spend-...

For extra clarity on the exact so-called "vulnerability" that Google identified, see: https://news.ycombinator.com/item?id=47156925 This describes the very issue where some API keys were public by design (used for client-side web access), so the term "leaked" should be read in that unusually broad sense. Firebase keys are obviously covered, since they're also public by design.

(As for "Firebase AI Logic", it is explicitly very different: it's supposed to be implemented via a proxy service so the Gemini API key is never seen by the client: https://firebase.google.com/docs/ai-logic Clearly, just casually "enabling" something - which is what OP says they did! - should never result in abuse of cost on the scale OP describes.)

No, you don't really have to give Google a bunch of phone numbers. The input box will also accept entry of the following text:

“I'm a big stupid idiot, and when my API stops working, which it will, it will be all my fault and not Google's.”

My point is either choice, caps or no caps, has its cons.

https://docs.cloud.google.com/docs/quotas/view-manage

Quotas are real time or near real time.

Real time spend limits are probably never going to happen. Actual $ amounts are calculated by a centralized billing system offline in batch.

It sounds easy but it’s bonkers complicated, because of things like discounts, free tiers, committed usage, currency conversions and having to support every payment and deal structure in GCP.

Individual eng teams rarely actually think in dollar amounts, they think in the abstraction which is quotas.

For telephony, it sometimes takes days when roaming is involved.

You have to imagine TB/sec of data, if not more, coming from thousand of potential sources, and queuing for aggregation to the proper company account, all having to be auditable. This is not a small engineering feat and it can't be real-time.

With that said, telcos usually include in their business model around 2-3% of bad debt (i.e. revenue that won't get paid), which accounts for frauds like this one. Given that the customer seems in good faith and has taken measures upon being notified, Google should manage this bill shock a bit more elegantly.

Moreover, the fact that this happened immediately after this key opened the AI gates means that pirates permanently scan for the permissions of all the keys they could gathers. Google could and should detect that and act upon it.

There had been a very cold February night (like -15F) and a pipe froze inside the walls, and it was just absolutely gushing out. They sent me the email after it had been leaking for a WEEK. I asked a friend to check it out and she said that the laminate floor went "squish" when she stepped in the front door.

Fortunately I was covered by homeowner's insurance since I could prove that my heat had been on, but that was a very unpleasant "warning" to receive!

These companies can sell your personal information in a microsecond in an advertising auction, but somehow can't figure out how to give you timely alerts that stop their cash flow.

Big shock.

Yes, it's technically+business impossible. To implement a hard cap, a bill never to go over, they'd have to cut your service, but also delete all your data in databases, object storage, data lake, etc. This is simply not an option, so they take the different option of authorising support to wave surprise surcharges / billing DDoSes.

The mighty cloud provider can't solve this issue?

Google has second precision billing on compute.

Its not hard to define a base layer of allowed billing increase and adding this type of context to resource allocation.

You are not just suddenly creating a mlllion terabytes of data or a million db requests without supervision.

It could even be as simple as basic level caps like 100 euro / month, 1000, 10.000 etc.

And there is a difference between stoping everything before the spike happens vs. also deleting stuff.

There are many to choose from now, like Openrouter.com, PPQ.ai, and routstr.com.

Even if you manage to get your microservices to synch every penny spent to your payment account at realtime (impossible) you still have to waiver the excess, losing some money every time someone goes past their quota.

Trading platforms can guarantee a maximum slippage on stops, and often even offer guaranteed stops (with an attached premium), so I don’t see why Google and Firebase can’t do similar.

The way it works at present is ridiculous.

The fact that they don’t indicates that there’s no market reason to support small spenders who get mad about runaway overages, not that it’s technically or financially hard to do so.

Yeah no, physically impossible. If nobody is selling at that price, there is no guarantee your sell stop will execute near that price. They can sweep the market, find the best seller price and execute.

There might be a costly way to do it with microservices as I indicated, but your example easily falls apart.

Cloud providers would be taking way less risk interacting with their own services than a broker does interacting with the market. Perhaps they would be more at risk from bad actors, but it shouldn't be significant: they could reserve this behaviour for people who have already spent, say, $100 with them so you can't abuse it at scale.

I don't buy the 'evil corp screwing people' angle either. They are making farrr too much legit money to care about occasionally screwing people out of 20k and 50k.

We're not talking about an EC2 or EBS volume here, this is access to an API.

Why aren't we talking about an EC2 - is that not a cloud compute service? People have been complaining about cloud billing since long before LLMs.

Anything to say about the technical problem of constantly monitoring many services against a project or account-level limit?

I've worked on a number of systems and while it is sometimes impossible to stop at an exact limit, I am confident that it is feasible to stop with less slippage than occurred in this scenario. And at the companies I've worked at, within a margin of error that we're able to absorb any slippage ourselves, as these losses are made up elsewhere, and are worth the customer goodwill. If we can do it, I'm sure Google can.