Hacker Newsnew | past | comments | ask | show | jobs | submit | lima's commentslogin

Yes it will, there's a clear purpose and the customer explicitly agrees.

Fable on GCP requires accepting a 60-day retention policy: https://cloud.google.com/terms/advanced-ai-safety-addendum

I don't think it mentions sharing the data with third parties such as Anthropic?


> Through Google Cloud's Agent Platform: Retention will need to be enabled for your new covered model, and retained data stays in your GCP environment. When models become available, onboarding details will be shared.

From https://support.claude.com/en/articles/15425996-data-retenti...


At least it stays in your GCP environment, AWS disclosure says that it will leave your data privacy and security boundary.

That Claude support page says the exact same thing about AWS (“retained data stays in your AWS environment”). AWS’s docs say differently, though, so it seems one of them has incorrect documentation. I wouldn’t necessarily trust the Claude docs to be correct even regarding GCP until some of this is ironed out.

edit: Google’s own docs also say zero data retention isn’t possible with Fable and your data will be retained for 60 days “outside of your account”. I’m doubtful that this data sharing is an AWS-only thing.


The data-sharing surely is for all providers. I think the sentence "When models become available, onboarding details will be shared." hides a lot of things.

Higher IO latencies in HDs might actually make this attack easier - more contention means more bits of data.

Shor published multiple quantum algorithms, including one for discrete logarithms. The term is sometimes used interchangeably.

They're closely related, ECC and RSA are both instances of the hidden subgroup problem.


> How do you know the clanker respects the instruction not to search the internet?

You can't, but given that it's a previously unsolved problem, it doesn't seem relevant? (nor are the author's potential biases - the claims are easily verified independently)


We tried this, but the quota for Opus models defaults to 0 on VertexAI and quota increase requests are auto-rejected.

Any tips?


What? There's no quota at all. You pay per token up to infinity.


There are in fact quotas and rate limits in VertexAI, albeit generous and automatically increased based on spend


No, unless you count tricks which are explicitly against ToS


They use a buffer battery, it's quite feasible with that.


Feels like such a waste for marginal gains?

With the range as good as a modern EV the charge time already isn't a particularly that bad. I'd much prefer more chargers (so that you can combine charging with something else you were going to do anyway) than faster ones.


I tend to agree but I think the strategy here is to convert people who stubbornly cling to gas vehicles because EVs somehow defy their expectations. I have been approached many times at highway rest stops by people who are curious and slightly skeptical about the EV value proposition. They see me hanging around the vehicle for a half hour and think “ugh, no thanks” as if that’s all I do when I travel. What they’re not seeing is that I rarely use public chargers at all, because 99% of my charging is done either at home or at the charger in the parking lot at work. It’s really just road trips. Not to mention, if you’re an ICE owner hanging around long enough at a rest stops to notice that I’m hanging around, are you really that much faster on a road trip?!!

Back on topic, I am ok with losing a little efficiency in the fast charging process if it means that more people switch away from a horribly inefficient and polluting technology.


You can still use OpenCode with the Anthropic API.


Yep. That's what I do. Just API keys and you can switch from Opus to GPT especially this week when Opus has been kind of wonky.


I pay $100/mo to Anthropic. Yesterday I coded one small feature via an API key by accident and it cost $6. At this rate, it will cost me $1000/mo to develop with Opus. I might as well code by hand, or switch to the $20 Codex plan, which will probably be more than enough.

I'd rather switch to OpenAI than give up my favorite harness.


Yeah I had a similar experience one time. Which is why I laugh when people suggest Anthropic is profitable. Sure, maybe if everyone does API pricing. Which they won’t because it’s so damn expensive. Another way to think about it is API pricing is a glimpse into the future when everyone is dependent on these services and the subscription model price increases start.


I don't get why people talk about ChatGPT as some great saviour though, they're in the same boat but just have more money to burn.


This is the intention. They do not want folks that can’t pay to use their service.


SOTA models cost SOTA prices. Nothing new there


Out of curiosity, what's your next monthly subscription in terms of price?


Electricity, $95/mo.


Now you got me thinking my electric company should start offering subscription tiers in these uncertain energy times...


Ours never will, they're a cartel, sadly. If you mean fixed subscription, next one is Netflix, I think, or my server provider at $40 or so.


My monthly "connection fee" is more than that (no solar, just EV). Your cartel needs to step it up!

For me it's $0.8/kWh during peak, $0.47 off peak, and super off peak of $0.15. I accidentally left a little mini 500W heater on all day, while I was out, costing > 5% of your whole month!


Wow, what the hell.


[flagged]


Yes, you are doing it too with antropic an xAI. I don't get your point. xAI and OpenAI are a little worst? Maybe, still very well fascism.


Quite a lot worse. Both OpenAI and xAI were among the largest donors of Trump's campaign

Musk was the largest individual political donor of the 2024 election [1] and Greg Brockman was the largest donor to Trump's "MAGA Inc" super PAC [2]

[1] https://www.washingtonpost.com/technology/2024/12/06/elon-mu...

[2] https://www.theverge.com/ai-artificial-intelligence/867947/o...


You’re right, Anthropic is quite a bit worse https://www.washingtonpost.com/technology/2026/03/04/anthrop...


Wait - are you missing all the context on this? Anthropic pushed back against this hard, there was a whole back and forth. I'm on mobile and can't look it up for you atm but if you google about this scenario, Anthropic definitely come out of this looking a lot better than OpenAI and xAI


Did you read the article? Or are you just replying?

Anthropic has literally been working with the DoD and Plantair for 2 years now. They were key to the Iran invasion.

If thats “looking better”, keep it.


Key being “worked”, past tense

Now they’re blacklisted from government work (appeal pending) and OpenAI practically jumped to replace them immediately


You’re right it completely doesn’t matter they’ve been instrumental to ICE and the war in Iran. It’s fine now, their previous actions are excused.

Edit: the word you’re also looking for is “working” not “worked”. I have many friends on government contracts still using Claude.


If you evaluate fascism in terms of donation, yes.

But it is more about the political opinions, IMHO, and Anthropic doesn't sound more attractive than the competitors. Anthropic is very much to the right of the transhumanism spectrum (even if xAI and OpenAI are even farther).


IMO, OpenAI have either implicitly committed to becoming the IT service for Trump's secret police, or they've willingly signed up for the harsh retaliation Anthropic's getting, knowing that the Trump administration will inevitably try to push OpenAI around in the same way, if they meaningfully refuse to assist in domestic mass surveillance efforts.


Anthropic was fine doing the same, they just didn't want it done to Americans.


OpenAI agreeing to operate as Trump's secret police materially impacts your security as a European, though, because it cements Trump's power.


Again, though, both of them agreed to that. Anthropic just didn't want to spy on Americans.


You can argue a moral equivalence, I guess, but on a practical level, OpenAI's decision is more dangerous for everyone, because it will help to secure Trump as a dictator.


Or have Claude write the code and Gemini review it. (Was using GPT for review until the recent Pentagon thing.)


You can also review the code you ship yourself.


I certainly do -- but having Gemini review it first saves a lot of time.


'just API key' lol. just hundreds of dollars at a minimum


Yes. And many companies pay that.


[flagged]


I'm testing glm5 on Claude code and opencode just to stop consuming American... Soo good so far!


Qwen works fine and requires paying no-one except a hardware vendor.


Fun fact: In Germany, the civil courts will usually take the case anyways if it has merit, but the winner ends up paying for the whole lawsuit if they failed to make an effort to resolve the case before suing.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: