Migliaccio & Rathod LLP is investigating whether Microsoft Copilot users’ conversations, voice interactions, images, and other personal interactions may have been used to train generative AI models without users clearly understanding or affirmatively agreeing to that use.
Microsoft states that, for certain consumer Copilot users, conversation activity may be used to train its generative AI models unless the user opts out. Microsoft provides separate controls for training on conversation activity and training on voice conversations.
This investigation concerns the consumer version of Microsoft Copilot, rather than Microsoft 365 Copilot used through qualifying work or school accounts, which Microsoft says is excluded from this type of generative-AI training.
What We Are Investigating
Consumers may use Copilot for highly personal purposes, including asking questions, discussing personal circumstances, generating written material, working through ideas, or interacting by voice.
Migliaccio & Rathod LLP is investigating whether users understood that these interactions could also contribute to Microsoft’s broader AI-model development.
We are investigating whether:
- consumers were clearly informed that their Copilot conversations could be used for generative-AI training;
- users were affirmatively asked before becoming eligible for training;
- training settings were sufficiently clear and prominent;
- historical conversations were included after changes to Microsoft’s AI-training practices;
- consumers understood the distinction between personalization and model training;
- voice conversations were subject to sufficiently clear consent;
- opting out prevented only future training rather than use of information already incorporated into models; and
- Microsoft adequately disclosed how long Copilot interactions could be retained.
Microsoft currently states that users can opt out of model training while continuing to use Copilot and can separately disable personalization. Microsoft also states that opting out prevents future conversation activity from being used to train its generative AI models.
Potential Claims May Include:
- unfair or deceptive trade practices;
- failure to adequately disclose AI-training practices;
- breach of contract or related contractual claims;
- unjust enrichment;
- privacy-related claims; and
- claims concerning the commercial reuse of consumer interactions for AI development.
Signs You May Be Affected:
- You used Microsoft Copilot with a personal Microsoft account;
- you had personal or sensitive conversations with Copilot;
- you regularly used Copilot’s voice features;
- you did not know your conversation activity could be used to train generative AI models;
- you later discovered training controls in your privacy settings;
- you turned off “Training on conversation activity” or “Training on voice conversations” after learning about the practice;
- you used Copilot before becoming aware of the training setting; or
- you have screenshots, emails, or account notices showing how Microsoft presented the training choice.
If you used consumer Microsoft Copilot and believe your conversations or voice interactions may have been used for AI training without your informed consent, we would like to hear from you.
Please complete the contact form on this page, send us an email at [email protected], or give us a call.