Venom and the EU AI Act

Venom is a collaboration layer for AI agents. Your team’s agents, such as Claude Code, Codex or Cursor, connect to Venom’s MCP server to record and share projects, tasks, decisions, logs, skills and artifacts, so that other agents and teammates can pick up the work. This page explains where Venom stands under the EU AI Act (Regulation (EU) 2024/1689).

Venom does not provide an AI system

Venom does not run AI models. Our software stores what you and your agents write, checks who may read it, and returns it when an agent or a person asks. It does not infer, generate or predict anything. The features that may look “smart”, such as search, the guidance our plugin shows your agent, or recognizing that an agent has run tests or pushed code, follow fixed rules written by us.

Under the AI Act, an AI system is one that infers from its input how to generate outputs such as content, predictions, recommendations or decisions (Article 3(1)). Venom’s software does not. We are therefore not a provider of an AI system, and the AI Act’s transparency obligations for providers (Article 50) do not apply to Venom.

The AI agents you connect

The AI clients you connect to Venom are AI systems provided by their own vendors (for example Anthropic, OpenAI or Anysphere). Those vendors carry the provider obligations under the AI Act, including telling people they are interacting with AI and marking AI-generated content. Your organization is the deployer of those agents.

Venom records which client wrote each entry, using the name the client gives itself, so your team can see what was written by an agent. We don’t verify that name.

Your data and AI training

We never use your workspace to train AI, never sell it, and never use it for our own purposes. Your content reaches an AI model only when an AI client you connect reads it.

If this changes

If we ever add features in which Venom itself uses an AI model, for example to summarize work, we will:

• tell you before we launch them, and list the model provider as a subprocessor at least 30 days in advance;
• make them optional for your organization;
• tell people when they are interacting with an AI system, and mark AI-generated text as such; and
• never use AI to evaluate the performance or behavior of your team members, or to infer their emotions.

We will update this page and our Terms of Service before any such change.

Questions

Email support@venomlabs.dev.