In other AI news, Reddit is asking AI companies to pay up, Meta is tagging real photos as "Made with AI", and a former Snap engineer has created a social network for AI called Butterflies. OpenAI rival Anthropic has released its newest generative AI model, Claude 3.5 Sonnet, and LinkedIn is offering AI-written cover letters. A recent study has looked at AI's political leanings, and Adobe has angered artists with new Photoshop terms.
Key takeaways:
- YouTube is making it easier to report deepfake videos that use a person's likeness without consent, treating it as a privacy issue rather than a content moderation issue.
- YouTube won't automatically take down these videos, but will consider factors like whether it's a parody or satire, whether the person making the request can be identified, or if it's spoofing a public figure.
- YouTube is also working on a process for removing deepfakes of copyrighted music, which have become increasingly common with tools like Suno and Udio.
- The success of this move will depend on how well enforced it is, and if YouTube can speed up the process of removing potentially harmful content.