Tokenless

Google deep mind

Stealing Reasoning Traces from Proprietary LLM APIs — Ilia Shumailov & Alexander Panfilov

Stealing Reasoning Traces from Proprietary LLM APIs — Ilia Shumailov & Alexander Panfilov

A paper by Ilia Shumailov and Alexander Panfilov exposes a critical vulnerability in proprietary LLM APIs: encrypted reasoning traces, returned for conversation state management, can be extracted and replayed. This enables universal jailbreaking, privacy leaks of sensitive user data, and poisoning of AI agent traces. The study highlights significant implications for AI safety, model monitorability due to opaque internal reasoning, and even subtle forms of "distillation" where smaller models mimic frontier ones. The discussion covers architectural and system-level defenses, advocating for rigorous scientific inquiry in AI safety research.

Don't Ship Skills Without Evals — Philipp Schmid, Google DeepMind

Don't Ship Skills Without Evals — Philipp Schmid, Google DeepMind

Philipp Schmid from Google DeepMind emphasizes the critical, often-overlooked need for rigorous evaluation of AI agent skills. He argues that shipping skills without testing is akin to deploying code without unit tests, leading to unreliable agent behavior. The talk covers what defines an agent skill, strategies for writing effective and correctly triggering skills, and a practical guide to building lightweight evaluation harnesses to catch failures proactively.

From Transcription to Live Music: Gemini's Audio Stack — Thor Schaeff, Google DeepMind

From Transcription to Live Music: Gemini's Audio Stack — Thor Schaeff, Google DeepMind

Thor Schaeff from Google DeepMind demos the advanced audio AI stack, starting with a single API call to Gemini for rich transcription (speaker names, emotions, translation). He showcases speech generation directed by "director's notes" instead of a voice catalog, the real-time, sound-to-sound Gemini 1.5 Flash Live model, and a live demo of Gemini Live using the Lyria 2 model as a tool to generate a full song on stage.

Any-to-Any: Building Native Multimodal Agents - Patrick Löber, Google DeepMind

Any-to-Any: Building Native Multimodal Agents - Patrick Löber, Google DeepMind

Patrick Löber from Google DeepMind provides a technical walkthrough of the Gemini API's "any-to-any" capabilities. The session covers multimodal understanding of complex documents, video, and audio; an agentic loop using function calling to trigger native image and speech generation; and the real-time, audio-to-audio Live API.

How to Build the Future: Demis Hassabis

How to Build the Future: Demis Hassabis

Demis Hassabis, CEO of Google DeepMind, outlines the remaining challenges on the path to AGI, including memory, continual learning, and true reasoning. He discusses how learnings from AlphaGo are shaping agent development, the strategic importance of powerful small models like Gemma, and his vision for AI as the ultimate tool for scientific discovery, offering a framework for identifying breakthrough opportunities and advice for founders building in the age of AI.

The arrival of AGI | Shane Legg (co-founder of DeepMind)

The arrival of AGI | Shane Legg (co-founder of DeepMind)

Shane Legg, Chief AGI Scientist at Google DeepMind, outlines his framework for AGI, predicting 'minimal AGI' within years and 'full AGI' within a decade. He details a path to more reliable systems and introduces 'System 2 Safety' for building ethical AI. Legg issues an urgent call for society to prepare for the massive economic and structural transformations that advanced AI will inevitably bring.

Next →