AI drafting and analysis tool

At Thomson Reuters, we do not use your user content and user prompts to train or improve CoCounsel Essentials or any third-party GenAI large-language models (LLMs), we do not use them in Thomson Reuters or third-party output, and neither OpenAI GPT nor Google Gemini stores them.

Thomson Reuters GenAI third-party partners, such as OpenAI and Google, are contractually prohibited from using any customer data to train their models. Additionally, where applicable, we have established systemic controls to turn off third parties' abuse monitoring solutions to ensure there is no human access to Thomson Reuters customers' data or inclusion of that data in third-party models.

Retention of user content and prompts is customized to fit customer needs and set up at the organization account level with the support of a Thomson Reuters client manager. We then delete user content and user prompts in accordance with the customer's retention policy. Users also have the ability to delete their content and prompts in the product at any time.

Thomson Reuters secures all data in transit via TLS 1.2, and symmetric encryption (AES-256) protects data at rest. All interactions with GenAI LLM providers use zero-retention API calls. All GenAI LLM requests originating from Thomson Reuters are processed under the identity "Thomson Reuters," never identifying the Thomson Reuters customer who made the request.

Thomson Reuters analyzes de-identified usage information to manage account health, plan for capacity, and develop our product roadmap.

Systematic procedures and hosting are in the U.S. We process all interactions with OpenAI GPT and Google Gemini in the U.S. We host backup data in separate availability time zones in the U.S.

Region-based hosting is available in the UK, Australia, and Canada.

Thomson Reuters has developed and implemented prompting guardrails, which ensure our GenAI-based products will only produce responses from content drawn from a clearly defined and limited set of verified information, such as our own content databases. This practice restricts the model's ability to produce potentially objectionable content, as it does not source information from the internet.

Our Trust Team of attorneys manually creates and peer reviews tests for CoCounsel that represent real-world legal work. After comparing CoCounsel’s results to their manually created ideal response, they train an LLM to do regular, automated reviews. Every night, CoCounsel completes more than 1,500 such tests under attorney oversight, and in total, we’ve run more than one million tests on CoCounsel since its launch. For an in-depth look at our testing process, read our blog post about benchmarking.

Any user content and prompts remain the user's property to the extent the user has such ownership rights. Subject to applicable law and Thomson Reuters terms and conditions, users own the outputs generated by CoCounsel Essentials.