Yes, CoCounsel can analyse and process documents in multiple languages, including but not limited to Japanese, French, German, Portuguese, and Spanish. This feature makes it a valuable tool for global legal practices, allowing legal professionals to access critical insights regardless of the language of the documents they are working with.

No. Thomson Reuters does not use your content or prompts to train or enhance CoCounsel or any third-party LLMs. Your information is not included in Thomson Reuters' or third-party outputs, and it is not stored by OpenAI GPT. Thomson Reuters AI third-party partner is 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 is configurable at the organisation level to align with each customer’s policy, set up with assistance from a Thomson Reuters client manager. We delete user content and prompts according to the customer’s retention schedule. Users can also delete their own content and prompts within the product at any time.

Data is encrypted in transit using TLS 1.2 and at rest with AES 256. All interactions with GenAI LLM providers use zero retention APIs, and requests are sent under the “Thomson Reuters” identity—never revealing the individual customer behind the request.

Thomson Reuters analyses de-identified usage metrics to monitor account health, forecast capacity needs, and inform the product roadmap.

Thomson Reuters recognise the importance of data residency for legal professionals and is committed to delivering AI-powered tools like CoCounsel in a secure, compliant, and locally relevant manner.

Yes, all customer prompts, content, and output are currently stored in Australia and will soon be processed in Australia as well, to better support our customers’ regulatory, compliance, and data sovereignty requirements. 

Thomson Reuters uses prompting guardrails that confine GenAI outputs to vetted sources, such as Thomson Reuters’ own content databases. By limiting the model to a defined set of verified information—and not the open internet—these controls help prevent biased or inappropriate responses.

A dedicated Trust Team of lawyers designs and peer reviews real world legal tests for CoCounsel. They compare CoCounsel’s outputs to expert-crafted answers and use those results to train an LLM for ongoing automated evaluations. Under lawyer oversight, CoCounsel runs over 1,500 tests nightly, with more than one million tests completed since launch.

For more details, see our benchmarking blog post.

Your content and prompts remain yours to the extent you hold ownership rights. Subject to applicable law and Thomson Reuters’ terms and conditions, you also own the outputs produced by CoCounsel.