Overview
- A joint statement says OpenAI strengthened Sora 2 guardrails after Bryan Cranston reported unauthorized generations of his voice and likeness during the app’s invite-only launch.
- OpenAI, SAG-AFTRA, CAA, UTA, and the Association of Talent Agents outlined an opt-in standard for likeness and voice, with OpenAI pledging to review complaints expeditiously and affirming that individuals control if they can be simulated.
- At the request of the Martin Luther King Jr. estate, OpenAI paused depictions of Dr. King and says it is tightening protections around historical figures.
- Following early backlash over copyrighted characters, OpenAI shifted from broad opt-out allowances to more granular controls, with the app now flagging or rejecting prompts that resemble third-party likenesses or IP.
- OpenAI added visible watermarks and invisible provenance signals to Sora videos, though reports show these can be stripped, and the company reiterated support for the proposed NO FAKES Act as the fast-growing app faces ongoing rights and enforcement challenges.