Migliaccio & Rathod LLP is investigating whether freelancers’ work product, proposals, messages, and other information submitted through Upwork may have been used to train Upwork’s artificial-intelligence models without sufficiently clear or informed consent.
Upwork maintains AI preference controls governing whether certain user information can be used to train and improve its AI models.
For freelancers, the information potentially involved can include past work, proposals, messaging activity, and project-related content. For clients, it may include project requirements, communications, and information concerning prior hiring activity. Upwork states that these materials can be used to develop more personalized AI-powered features.
The use of freelancer work product is particularly significant because users may have created that material specifically for paying clients rather than for development of Upwork’s broader AI systems.
What We Are Investigating
Migliaccio & Rathod LLP is investigating:
- when Upwork first began using user information to train its AI systems;
- what freelancers and clients were told before that use began;
- whether historical communications or work product became eligible for training;
- how Upwork’s AI preference settings were initially configured;
- whether users affirmatively agreed before their work or communications were used;
- whether freelancers understood that commercially valuable work product could contribute to Upwork’s AI models;
- what happens to content already used when a user later opts out; and
- whether users received adequate compensation or other value for this additional use of their work.
Upwork currently states that when a freelancer and client communicate or exchange work product, both parties must be opted in before that joint communication or work product can be used for this type of AI training.
However, Upwork also states that if a user was previously opted in and later opts out, content shared during the opted-in period may continue to be used to train Upwork’s models. New material shared after the opt-out is excluded.
That raises questions about whether users can meaningfully withdraw work product or communications after they have already entered the training process.
Potential Claims May Include:
- unfair or deceptive trade practices;
- breach of contract;
- breach of the implied covenant of good faith and fair dealing;
- unjust enrichment;
- inadequate disclosure concerning the reuse of freelancer work product; and
- claims concerning the commercial use of user-created material for AI development.
Signs You May Be Affected:
- You worked as a freelancer or client on Upwork;
- you submitted proposals, writing, designs, code, images, reports, or other work product through the platform;
- you exchanged substantial project-related communications through Upwork;
- you did not know that your information could be used to train Upwork AI models;
- you later discovered an AI preferences setting in your account;
- you changed your preference from allowing training to opting out;
- you learned that earlier content could remain in models after opting out;
- you used Upwork before the current AI preference controls were introduced; or
- you have older screenshots, agreements, emails, or privacy notices concerning how your work product could be used.
Upwork currently states that third-party technology providers used to power its AI features are not permitted to train their own models on Upwork customer data. This investigation concerns Upwork’s use of user information to develop and improve its own AI-powered systems.
If you provided work product or communications through Upwork and believe that material 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.