Anthropic By The Bay

This year at By The Bay, we’re thrilled to be joined by Anthropic, a company at the forefront of building large-scale AI systems with safety and reliability as first principles— not afterthoughts.

Representing them is Vinay Rao, former Head of Safeguards and currently Advisor to the company. Vinay has led safety and trust teams at YouTube, Stripe, Airbnb, and Google, building infrastructure and methods to evaluate and mitigate real-world AI failure modes. Vinay will presents "Safety in Foundational Models," a technical overview of why current evals miss key failure modes, how multi-modal systems break under shift, and what it takes to build scalable, proactive safeguards.

RSVP

What Is Anthropic?

Founded in 2021 by a group of former OpenAI researchers, Anthropic set out with an unusually clear mission: to build AI systems that are not just advanced, but aligned with human values from the ground up.

As a Public Benefit Corporation, they’ve anchored their work in the principle that AI should be helpful, honest, and harmless, a north star that guides everything from product design to research priorities. That mission now takes shape most visibly in the Claude 4 family—Opus 4.1 for top-tier capability, Sonnet 4 for balanced performance, and Haiku 3.5 for speed and efficiency—models that can process both text and images, and integrate seamlessly into apps via web and API. Underpinning these systems is Constitutional AI, a training framework that encodes ethical principles directly into the model’s behavior.

At the same time, Anthropic is pushing the boundaries of interpretability, using techniques like dictionary learning to uncover the concepts their models store and the reasoning chains they follow. Their current focus extends beyond pure research into enterprise-ready solutions, from AI-generated pre-meeting summaries to experiments with voice-driven assistants in collaboration with Amazon and ElevenLabs. They’re also investing in policy and economic foresight, launching tools like the Anthropic Economic Index and the Economic Futures Program to understand—and help society adapt to—the real-world impact of AI.