from Hacker News

Tell HN: Using Anthropic is risky. They just banned our company without warning

by ent101 on 11/22/24, 12:52 AM with 27 comments

• by airbreather on 11/22/24, 6:04 AM

  This is why the "cloud" model of AI is not really where I would want to be as a business.

  Admittedly you are not going to get the same cutting edge performance, but I would only be basing a business on an AI I could download and run on my own hardware, such that at least current performance could be guaranteed going forward.

  Anytime you buy anything as a service, you are in danger that due to malice, incompetence, lack of profitability, competing with a greater or established interest, or many other reasons, that service may become unavailable at little to no notice.

  I don't understand why companies put themselves at these risks, when it comes to software as a service components to their business, when they wouldn't dream of exposing themselves to similar risks in any other domain related to their well being.

• by bigfatkitten on 11/22/24, 3:45 AM

  The moral of this story:

  If your business relies on another business to provide a service, you either need an iron clad guarantee that they will continue to do so, or you need a plan B for when they don't.

• by talldayo on 11/22/24, 12:56 AM

  > we have no recourse but to cry for help on HN.

  HN doesn't wave a magic wand and make problems go away. You solve this by not building a business on an arbitrary service that can be revoked at-will by the provider.

  This isn't an "Anthropic is risky" problem - this is a "don't build your castle in other people's kingdoms" problem: https://howtomarketagame.com/2021/11/01/dont-build-your-cast...

• by 1123581321 on 11/22/24, 8:48 PM

  That's frustrating. I'm not saying you did anything wrong, but I wouldn't rely on any of these companies without talking to a human first. That's because they are still figuring out what acceptable use looks like, let alone how to describe it to customers, build dispute mechanisms etc.

  We use an Anthropic competitor and made sure we had a responsive account rep before committing to anything, and we're still rolling out in phases. We're moving quickly but we're not startup speed, so we have more time and margin to have those conversations.

• by readyplayernull on 11/22/24, 7:28 PM

  Anthropic is receiving large investments from Amazon, so it wouldn't surprise me whether they are "optimizing" their business model:

  https://news.ycombinator.com/item?id=42215126

• by tmaly on 11/25/24, 11:24 AM

  Can you not use the AWS Bedrock and still use the Anthropic models there without having to deal directly with them?

• by elmerfud on 11/22/24, 1:44 AM

  The typical responses you'll get here are summed to as "sucks to be you". The attitude of allowing larges places to have opaque processes for termination of a business relationship is staunchly defended here. To the point that you have no idea if it was even legally allowed for them to terminate the relationship. Just because you're a business doesn't mean you can absolutely refuse service for any reason at all. I also find it kind of funny that the same people would largely not grant the same rights to someone who refuses to decorate a cake in Colorado. Even though there's likely hundreds of other cake shops and decorators they could go to.

  This is one of those situations where most people will never care or they will blame you until it happens to them. This is absolutely something we need the legislation or at least the FTC to be involved with to greatly curtail these terms of service based things. Especially if you're in a business to business relationship this should be governed by a contract and not a one-sided terms of service which may or may not be legal and you had zero negotiating power in. With a contract there is actual specific provisions and methods required to terminate the contract early. Those things must be documented and defensible but terms of service is just good luck.

  The most laughable but also the most common argument in defense of these one-sided terms of service that I see from hacker news people is if they tell you exactly what you did wrong and gave you a chance to remediate it then you could work the system and not do that anymore. I thought that was the whole idea if you know absolutely what is right and what is wrong you can avoid doing what is wrong. But apparently having a guessing game what is appropriate and why it is appropriate and what is not and why it's not is par for the course for the vast majority of hacker news people.

  Is this a business or is this a dating relationship?

• by vednig on 11/22/24, 11:55 PM

  That's why rule one of Reliable Software keep Dependencies Low