Grok and non-consensual deepfakes: a test of AI product liability

The debate around Grok on X reopens who is liable for harmful generative AI outputs. In 2026, policy text is not enough—product guardrails decide outcomes.