Anthropic's new think tank

Good morning! Caution: the contents of this email may be hot! 🔥

This is the last day to enter the FryAI Bracket Challenge! It’s totally FREE to participate.

Compete against your fellow FryAI readers and ChatGPT, and win prizes!

👉 Fill out your bracket now!

🎁 Prizes:

What’s up? Anthropic has launched a new research group called The Anthropic Institute to study the major risks and opportunities that powerful AI could create for society.

Want the details? The Anthropic Institute will bring together researchers from fields like machine learning, economics, and social science to better understand how advanced AI could affect the world. The group will study issues such as job disruption, economic changes, cybersecurity risks, and how AI systems should be governed. It combines several of Anthropic’s existing research teams, including groups that test the limits of AI systems, study how AI is used in the real world, and track its impact on the economy. Led by Anthropic co-founder Jack Clark, the institute will also work with outside researchers, industries, and communities to share what it learns about rapidly advancing AI.

Why is this significant? AI systems are becoming more powerful very quickly, and companies like Anthropic believe the pace of progress is accelerating. That means society may soon face major decisions about how AI should be used, regulated, and aligned with human values. By studying these issues early and sharing information publicly, the institute hopes to help governments, businesses, and the public prepare for the benefits and risks of the next generation of AI.

What’s up? Grammarly says it is shutting down an AI feature that mimicked the writing style of real experts without their permission after facing criticism from writers and journalists.

Want the details? The feature, called Expert Review, used AI to suggest edits that were described as being “inspired by” the writing of well-known authors and commentators. It relied on publicly available writing and large language models to generate suggestions that sounded like advice from those experts. However, several writers said the tool misrepresented their voices and used their identities without consent. In response, Grammarly disabled the feature and said it will redesign it so experts can decide whether they want to participate and how their knowledge is used.

Why is this important? This controversy highlights a growing issue in AI: who owns a person’s voice, ideas, or style when machines can copy them. As AI tools become more common in writing and creative work, companies will face increasing pressure to get permission and give people control over how their expertise or identity is used.

Hi — I’m Hunter, a PhD candidate whose work has appeared in major academic journals and popular tech outlets. I founded FryAI to make staying ahead of AI clear, accessible, and fun.