On 2026-02-17 20:52, Seth via Boost wrote:
... The problem arises as soon copyrighted material is used during training. Yes, it does lead to interesting grey areas, after all humans do similar aggregation in practice. However, there's a very big difference in effort and implied merit involved.
To be frank an optimistic (?!) statement like "no new legislation required" seems to fly in the face of public opinion and common sense. ...
I am not a lawyer, but infringement is a real problem, with software, hardware as well as AI models. For IBM Granite models https://www.ibm.com/granite from "Read the IBM statement on IP protection" section near bottom of page: "... Moreover, contrary to some other providers of large language models and consistent with the IBM standard approach on indemnification, IBM does not require its customers to indemnify IBM for a customer's use of IBM-developed models. Also, consistent with the IBM approach to its indemnification obligation, IBM does not cap its indemnification liability for the IBM-developed models. ..." I don't know of any other AI model provider that provides a copyright indemnity for its models. If a third party sues you claiming that an IBM-developed AI model or its output infringes on their copyright, IBM will step in to defend you and cover the legal costs. Disclaimer: While I just finished my first semester as Student of Mathematics at age of 60 in early retirement, I am still IBM employee until age of 63. Regards, Hermann Stamm-Wilbrandt.