Loading issue research
Base-model developers should be liable by default; fine-tuners and deployers should be liable for substantial modifications or intentional misuse.
Verification Status
AI-researched, unverifiedLast Reviewed
Jul 4, 2026
Cited Sources
8
What is failing, what we would change, and the conclusion we are willing to defend.
The honest starting point here is that this question is nearly untested in court. Two teenagers died; two wrongful-death suits followed. Both name a company that is simultaneously the base-model developer and the deployer, so neither case presents the scenario this issue is about: a harm caused by an independent third party fine-tuning or wrapping someone else's base model. Both suits ended in settlement. No admission of liability. No ruling on who was at fault. That gap is the reason this issue exists as a distinct question from AI-02's frontier-safety framework, which regulates the base-model layer but doesn't say who's liable once a downstream party has customized what gets deployed.
Support federal legislation establishing base-model developer liability by default, with fine-tuners and deployers liable specifically for substantial modifications or intentional misuse through a defined allocation that avoids re-litigating the framework in every case.
Use a compute-based "substantial modification" threshold to make that line administrable, consistent with how this platform already uses compute thresholds elsewhere (AI-02, AI-06).
Treat AI-generated output as not covered by Section 230's third-party-content immunity, consistent with that statute's own authors' stated view. This doesn't decide who's liable; ordinary product-liability and negligence law decides the question once a 230 threshold dismissal is unavailable.
Support a rebuttable "stewardship defense" for base-model providers who can document safety controls and cooperation with downstream fine-tuners, an incentive to build safer base models. Compliance boxes alone provide no liability shield.
Revisit this position as case law develops. Almost no court has yet allocated fault between a base-model developer and an independent third-party fine-tuner in a final judgment.
Base-model developers should be liable by default; fine-tuners and deployers should be liable for substantial modifications or intentional misuse.
Turn frustration into useful pressure.
If this position misses evidence or a lived consequence, challenge it. If it holds up, help test it locally and connect it to the issues around it.