The OpenAI board was right
OpenAI’s use of a chatbot voice widely perceived as mimicking Scarlett Johansson’s performance in the film *Her* has triggered backlash over consent, likeness rights, and corporate ethics. Commenters debate whether hiring a soundalike (after Johansson reportedly declined) and promoting it with a “her” reference crosses legal lines like passing off and voice misappropriation, or is merely unethical but allowed. The controversy broadens into criticism of OpenAI’s data practices, its shift from nonprofit ideals to profit-seeking, and the concentration of AI power in the hands of leaders seen as manipulative or untrustworthy.
Alleged Scarlett Johansson Voice Misuse
- Many see OpenAI’s “Sky” voice as an intentional imitation of Johansson’s AI role in Her, especially given prior outreach to her and the CEO’s “her” tweet.
- Others argue the voice is clearly a different actor and just a generic “flirty American woman”; similarity is inevitable among finite human voices.
- Some think pulling the voice suggests legal worry or bad PR; others say removing it is not an admission of guilt.
Legal Debates: Likeness, Voice, and “Passing Off”
- Multiple comments cite right-of-publicity and voice-misappropriation precedents (e.g., Bette Midler, Tom Waits), arguing that using a soundalike to evoke a specific celebrity and role can be unlawful “passing off.”
- Skeptics question how to define “similar enough,” whether voice can be owned, and warn about slippery slopes for voice actors who naturally resemble famous voices.
- Several note that intent and marketing context (references to Her, timing of permission requests) would be central if litigated; outcome is seen as uncertain.
Data “Theft” and AI Training
- Strong sentiment that OpenAI—and big tech generally—are “built on theft,” repurposing decades of public writing and media without consent or royalties.
- Others counter that using public information for training is akin to humans learning from what they read; distinguishing human learning from large-scale commercial “lossy compression” is contested.
- Comparisons are made to search engines and cable TV; critics say LLMs, unlike search, don’t drive users back to original sources.
Culture, Governance, and Trust
- Many portray the CEO as power- and PR-driven, emblematic of a tech culture that treats “no” as negotiable and normalizes “move fast and break things.”
- Some defend his effectiveness at organizing teams and creating value, while still acknowledging pervasive dishonesty and consent problems.
- The shift from non-profit, “benefit humanity” origins to closed, for-profit behavior intensifies distrust and calls for regulatory or legal intervention.
LLM Value vs. Harm
- Engineers and users describe LLMs as impressive and useful for brainstorming, orientation in new domains, and “where do I start?” moments.
- Others say benefits are overhyped and likely outweighed by harms: content appropriation, energy and water use, labor displacement, and erosion of creative incentives.