7 ms·

Legal Contracts Built for AI Agents

72 points by arnon 1y ago

ataha322 1y ago
The question isn't just who's liable - it's whether traditional contract structures can even keep up with systems that learn and change behavior over time. Wonder if this becomes a bigger moat than the AI.
- tuesdaynight 1y ago
  Probably a dumb question, but what do you mean with changing behavior over time? Contract with changing clauses? From my limited knowledge on the matter, the idea of a contract is getting rules that would not change without agreement from both parties.
  - candiddevmike 1y ago
    I encounter this all the time with GenAI projects. The idea of stability and "frozen" just doesn't exist with hosted models IMO. You can't bet that the model you're using will have the exact behavior a year from now, hell maybe not even 3 months. The model providers seem to be constantly tweaking things behind the scenes, or sunsetting old models very rapidly. Its a constant struggle of re-evaluating the results and tweaking prompts to stay on the treadmill. Good for consultants, maybe, horrible for businesses that want to mark things as "done" and move them to limited maintenance/care and feeding teams. You're going to be dedicating senior folks to the project indefinitely.
    - htrp 1y ago
      You're gonna have to own the model weights and there will be an entire series of providers dedicated to maintaining oldmodels.
    - hodgesrm 1y ago
      This is a big motivation for running your own models locally. OpenAI's move to deprecate older models was an eye-opener to some but also typical behavior of the SaaS "we don't have any versions" style of deployment. [0] It will need to change for AI apps to go mainstream in many enterprises. [0] https://simonwillison.net/2025/Aug/8/surprise-deprecation-of-gpt-4o/ https://simonwillison.net/2025/Aug/8/surprise-deprecation-of...
    - idiotsecant 1y ago
      This isn't a new problem. It's like if you built a business based on providing an interface to a google product 10 years ago and google deleted the product. The answer is you don't sell permanent access to something you don't own. Period.
  - avs733 1y ago
    I interpreted the comment as worrying about drift across many contracts not one contract changing. Imagine I create a new agreement with a customer once a week. I’m no lawyer so might not notice the impact of small wording changes on the meaning or interpretation of each sequential contract. Can I try and prompt engineer this out? Yeah sure. Do I as a non lawyer know I have fixed it - not to a high level of confidence.
- bryanrasmussen 1y ago
  humans. Also it might be that with systems that learn and change behavior over time, some sort of contract structure is needed. Not sure if traditional is the answer though.
- lazide 1y ago
  You literally don’t want contracts that ‘learn and change behavior over time’? What is the stated use case here?
  - hodgesrm 1y ago
    No, at least not in all cases. Customers incur review costs and potentially new risks if you change contract terms unexpectedly. In my business many large customers will only adopt our ToS if we commit to it as a contract that does not change except by mutual agreement. This is pretty standard behavior.
    - lazide 1y ago
      I can’t think of any case where someone who cares about the contract (aka actual terms) would be okay with it just changing. Arguably, it violates the concept of a contract which in most legal systems requires a meeting of the minds. Do you have any examples where it would be okay?
      - hodgesrm 1y ago
        Only a fraction of our customers insist on locking the contract terms. It's far less than half, and it's not correlated in an obvious way to the value of the contract.
n8m8 1y ago
Can't scroll, Cookies disclaimer doesn't work in firefox with ublock origin :(
- aleatorianator 1y ago
  reader mode?
- Neywiny 1y ago
  That's why I always incognito. Sure, I accept your cookies. They're gone in a few hours anyway
Neywiny 1y ago
I'm not sure I understand why this is about agents. This feels more like contracting than SaaS. If I contract a company to build a house and it's upside down, I don't care if it was a robot that made the call, it's that company's fault not mine. I often write electronic hardware test automation code and my goodness if my code sets the power supply to 5000V instead of 5.000V (made up example), that's my fault. It's not the code's fault or the power supply's fault. So, why would you use a SaaS contract for an agent in the first place? It should be like a subcontractor. I pay you to send 10k emails a day to all my clients. If you use an agent and it messes up, that's on you. If you use an agent and it saves you time, you get the reward.
- nemomarx 1y ago
  To have that you need a human to take responsibility somewhere, right? I think people want to assign responsibility to the "agent" to wash their hands in various ways. I can't see it working though
  - arnon 1y ago
    If I am a company that builds agents, and I sell it to someone. Then, that someone loses money because this agent did something it wasn't supposed to: who's responsible? Me as the person who sold it? OpenAI who I use below? Anthropic who performs some of the work too? My customer responsible themselves? These are questions that classic contracts don't usually cover because things tend to be more deterministic with static code.
    - Xylakant 1y ago
      > These are questions that classic contracts don't usually cover because things tend to be more deterministic with static code. Why? You have a delivery and you entered into some guarantees as part of the contract. Whether you use an agent, or roll a dice - you are responsible for upholding the guarantees you entered into as part of the contract. If you want to offload that guarantee, then you need to state it in the contract. Basically, what the MIT Licenses do: "No guarantees, not even fitness for purpose". Whether someone is willing to pay for something where you enter no liability for anything is an open question.
jrm4 1y ago
Sigh -- another not-even-thinly-veiled ducking of "A computer can never be held accountable, therefore a computer must never make a management decision." This is not the way we want to be going.
- binarysneaker 1y ago
  Which way should we be going?
  - jrm4 1y ago
    More accountability for humans and/or corporations, not less?
Animats 1y ago
Legal contracts built for sellers of AI agents. The contract establishes that your agent functions as a sophisticated tool, not an autonomous employee. When a customer's agent books 500 meetings with the wrong prospect list, the answer to "who approved that?" cannot be "the AI decided." It has to be "the customer deployed the agent with these parameters and maintained oversight responsibility." The MSA includes explicit language in Section 1.2 that protects you from liability for autonomous decisions while clarifying customer responsibility. The alternative is that the service has financial responsibility for its mistakes. This is the norm in the gambling industry. Back when GTech was publicly held, their financial statements listed how much they paid out for their errors. It was about 3%-5% of revenue. Since this kind of product is sold via large scale B2B deals, buyers can negotiate. Perhaps service responsibility for errors backed up by reinsurance above some limit.
nadis 1y ago
> "The template uses CommonPaper's Software Licensing Agreement and AI Addendum as a foundation, adapted for the unique characteristics of AI agents. Nick and the GitLaw team built this based on patterns from reviewing hundreds of agent contracts. We contributed our research from working with dozens of agent companies on monetization challenges." Unless I'm misunderstanding and GitLaw and CommonPaper are related or collaborating, I feel like this callout deserves to be mentioned earlier on and the changes / distinctions ought to be called out more explicitly. Otherwise, why not just use CommonPaper's version?

72 points by arnon 1y ago

The question isn't just who's liable - it's whether traditional contract structures can even keep up with systems that learn and change behavior over time. Wonder if this becomes a bigger moat than the AI.

tuesdaynight 1y ago
Probably a dumb question, but what do you mean with changing behavior over time? Contract with changing clauses? From my limited knowledge on the matter, the idea of a contract is getting rules that would not change without agreement from both parties.
- candiddevmike 1y ago
  I encounter this all the time with GenAI projects. The idea of stability and "frozen" just doesn't exist with hosted models IMO. You can't bet that the model you're using will have the exact behavior a year from now, hell maybe not even 3 months. The model providers seem to be constantly tweaking things behind the scenes, or sunsetting old models very rapidly. Its a constant struggle of re-evaluating the results and tweaking prompts to stay on the treadmill. Good for consultants, maybe, horrible for businesses that want to mark things as "done" and move them to limited maintenance/care and feeding teams. You're going to be dedicating senior folks to the project indefinitely.
  - htrp 1y ago
    You're gonna have to own the model weights and there will be an entire series of providers dedicated to maintaining oldmodels.
  - hodgesrm 1y ago
    This is a big motivation for running your own models locally. OpenAI's move to deprecate older models was an eye-opener to some but also typical behavior of the SaaS "we don't have any versions" style of deployment. [0] It will need to change for AI apps to go mainstream in many enterprises. [0] https://simonwillison.net/2025/Aug/8/surprise-deprecation-of-gpt-4o/ https://simonwillison.net/2025/Aug/8/surprise-deprecation-of...
  - idiotsecant 1y ago
    This isn't a new problem. It's like if you built a business based on providing an interface to a google product 10 years ago and google deleted the product. The answer is you don't sell permanent access to something you don't own. Period.
- avs733 1y ago
  I interpreted the comment as worrying about drift across many contracts not one contract changing. Imagine I create a new agreement with a customer once a week. I’m no lawyer so might not notice the impact of small wording changes on the meaning or interpretation of each sequential contract. Can I try and prompt engineer this out? Yeah sure. Do I as a non lawyer know I have fixed it - not to a high level of confidence.
bryanrasmussen 1y ago
humans. Also it might be that with systems that learn and change behavior over time, some sort of contract structure is needed. Not sure if traditional is the answer though.
lazide 1y ago
You literally don’t want contracts that ‘learn and change behavior over time’? What is the stated use case here?
- hodgesrm 1y ago
  No, at least not in all cases. Customers incur review costs and potentially new risks if you change contract terms unexpectedly. In my business many large customers will only adopt our ToS if we commit to it as a contract that does not change except by mutual agreement. This is pretty standard behavior.
  - lazide 1y ago
    I can’t think of any case where someone who cares about the contract (aka actual terms) would be okay with it just changing. Arguably, it violates the concept of a contract which in most legal systems requires a meeting of the minds. Do you have any examples where it would be okay?
    - hodgesrm 1y ago
      Only a fraction of our customers insist on locking the contract terms. It's far less than half, and it's not correlated in an obvious way to the value of the contract.

n8m8 1y ago

Can't scroll, Cookies disclaimer doesn't work in firefox with ublock origin :(

aleatorianator 1y ago
reader mode?
Neywiny 1y ago
That's why I always incognito. Sure, I accept your cookies. They're gone in a few hours anyway

Neywiny 1y ago

I'm not sure I understand why this is about agents. This feels more like contracting than SaaS. If I contract a company to build a house and it's upside down, I don't care if it was a robot that made the call, it's that company's fault not mine. I often write electronic hardware test automation code and my goodness if my code sets the power supply to 5000V instead of 5.000V (made up example), that's my fault. It's not the code's fault or the power supply's fault. So, why would you use a SaaS contract for an agent in the first place? It should be like a subcontractor. I pay you to send 10k emails a day to all my clients. If you use an agent and it messes up, that's on you. If you use an agent and it saves you time, you get the reward.

nemomarx 1y ago
To have that you need a human to take responsibility somewhere, right? I think people want to assign responsibility to the "agent" to wash their hands in various ways. I can't see it working though
- arnon 1y ago
  If I am a company that builds agents, and I sell it to someone. Then, that someone loses money because this agent did something it wasn't supposed to: who's responsible? Me as the person who sold it? OpenAI who I use below? Anthropic who performs some of the work too? My customer responsible themselves? These are questions that classic contracts don't usually cover because things tend to be more deterministic with static code.
  - Xylakant 1y ago
    > These are questions that classic contracts don't usually cover because things tend to be more deterministic with static code. Why? You have a delivery and you entered into some guarantees as part of the contract. Whether you use an agent, or roll a dice - you are responsible for upholding the guarantees you entered into as part of the contract. If you want to offload that guarantee, then you need to state it in the contract. Basically, what the MIT Licenses do: "No guarantees, not even fitness for purpose". Whether someone is willing to pay for something where you enter no liability for anything is an open question.

jrm4 1y ago

Sigh -- another not-even-thinly-veiled ducking of "A computer can never be held accountable, therefore a computer must never make a management decision." This is not the way we want to be going.

binarysneaker 1y ago
Which way should we be going?
- jrm4 1y ago
  More accountability for humans and/or corporations, not less?

Animats 1y ago

Legal contracts built for sellers of AI agents. The contract establishes that your agent functions as a sophisticated tool, not an autonomous employee. When a customer's agent books 500 meetings with the wrong prospect list, the answer to "who approved that?" cannot be "the AI decided." It has to be "the customer deployed the agent with these parameters and maintained oversight responsibility." The MSA includes explicit language in Section 1.2 that protects you from liability for autonomous decisions while clarifying customer responsibility. The alternative is that the service has financial responsibility for its mistakes. This is the norm in the gambling industry. Back when GTech was publicly held, their financial statements listed how much they paid out for their errors. It was about 3%-5% of revenue. Since this kind of product is sold via large scale B2B deals, buyers can negotiate. Perhaps service responsibility for errors backed up by reinsurance above some limit.

nadis 1y ago

> "The template uses CommonPaper's Software Licensing Agreement and AI Addendum as a foundation, adapted for the unique characteristics of AI agents. Nick and the GitLaw team built this based on patterns from reviewing hundreds of agent contracts. We contributed our research from working with dozens of agent companies on monetization challenges." Unless I'm misunderstanding and GitLaw and CommonPaper are related or collaborating, I feel like this callout deserves to be mentioned earlier on and the changes / distinctions ought to be called out more explicitly. Otherwise, why not just use CommonPaper's version?