TOP LAUNCHES

CapCut’s AI content generator, an AI research assistant for essays, and more.

Top launches

CapCut Commerce Pro is an AI-powered content production platform designed for e-commerce marketing. It lets you generate shoppable video ads, product images, and social content — all from a product link.

SWE-Kit is a headless IDE that comes packaged with a bunch of native AI tools that allow you to build your own custom coding agents, sort of like Cursor or Devin. 

PaperGen is an AI-powered tool that helps you generate well-structured long-form papers with fully referenced citations. It handles the research for you and presents a pretty convincing paper that is apparently AI checker-proof. 

Sona lets you turn your conversations into valuable insights. Record, transcribe, summarize, and chat with 99% accuracy in 99+ languages. It works with meetings, lectures, interviews, and more. 

Melies is an AI filmmaking platform. It takes your idea via natural language and turns it into a script. From there you can feed the script back into the platform and it will generate a movie complete with different scenes, characters, and visuals.

THE BIG IDEA

Will OpenAI kill Google’s search monopoly?

Rights for robots? Per a new report in Transformer, Anthropic recently hired its first “AI welfare” researcher, tasked with investigating whether models might become “morally relevant” agents in the future. In other words: Do the robots deserve rights? Are the chatbots sentient? Do they have interests that warrant protection, like humans? Will they eventually? How should we know?

These might sound like sci-fi questions, or hypotheticals from a philosophy seminar run amok, but some AI researchers believe they’ll become increasingly urgent as models improve. Anthropic’s new hire, Kyle Fish, recently co-authored a research paper arguing that we need to start assessing AI systems for evidence of consciousness and preparing policies “for treating [them] with an appropriate level of moral concern.”

That is, don’t harm the robots — if the robots are actually capable of perceiving harm (tbd). 

The paper doesn’t go into much detail about what these harm-reduction policies would look like, other than recommending that top AI companies hire AI welfare researchers to start studying the question. Our take? Right now, “AI welfare” remains the (near) exclusive concern of niche grad seminars and a handful of well-paid consultants. It’s fairly clear that current models don’t meet usual standards for “moral relevance” (e.g. sentience, capacity to experience pain, etc). But keep a close eye on this space — it will probably be the site of many bitter regulatory battles to come. — Sanjana

Overheard in the discourse

From a recent interview between Y Combinator CEO Garry Tan and OpenAI Founder Sam Altman:

Garry Tan: “What are you most excited about in 2025?”

Sam Altman: “AGI. I’m excited for that.”

…the singularity approaches?