According to heise online, Anthropic is reportedly consulting law firms. The question is how AI agents should negotiate terms, salaries and contracts for their users in the future. The report says nothing more, so it is better read as a sign of direction than as an announcement.

What is meant by AI agents

An AI agent is a program that does not just answer questions but carries out steps on its own. For example, it can gather information, compare offers and draft messages. In a negotiation, such an agent would act on behalf of a person and try to achieve a good outcome.

This raises questions that a chatbot has so far barely touched. May an agent make binding commitments? Who is liable if it promises something wrong? And how far does its authority extend if the user has only given a rough goal?

Why lawyers come into play

Negotiations are governed by law. Statements can create contracts, and deceptive claims can have consequences. When an agent acts in such situations, its developers need guardrails. They have to define what it may say, when it must ask for clarification and when a human makes the decision. Advice from law firms could help align such rules with existing law. Whether and how Anthropic implements this is not clear from the report.

More on this topic:

What readers can take away

Anyone who wants to use agents for negotiations in the future should first clarify what powers the tool has. It makes sense to tie binding agreements to an explicit confirmation. It is also worth looking at which personal data the agent discloses, such as salary expectations or budget limits.

An agent may also be sitting on the other side. Then two programs negotiate with each other, and clear rules become even more important. Until more details are made public, the report remains a signal that providers take the topic seriously and want to include legal questions early on.