AI #169: New Knowledge
via Substack Zvi [999] — Even in a relatively quiet period, AI is out there creating new knowledge.
Read
Welcome backWhat changed since your last visit
Current briefing
What changed today
Checking the latest coverage…
Building the latest development brief…
Open Situation Room →
Latest reporting
Updated hourly
The Case for Evaluating Model Behaviors
via Alignment Forum [999] — Most evaluations of AI systems focus on their capabilities: how good they are at coding tasks, how effectively they can answer complex scientific questions, and so on.From a safety perspective, capability evaluations have a place: by understanding how…
Read
You can now remix other people’s YouTube Shorts with AI
via The Verge AI [4] — Google announced a new YouTube Shorts Remix feature that lets users restyle clips or even insert themselves into other people's videos using Gemini Omni. Now, at the bottom of a YouTube Short, when you click the remix icon, you'll see an option to…
Read
Demis Hassabis said this might be the ‘foothills of the singularity.’ What?
via The Verge AI [6] — Welcome to a "profound moment for humanity," according to Google DeepMind CEO Demis Hassabis, who closed out Google I/O's keynote presentation on Tuesday, saying: Google's cutting-edge research and products will help unlock AGI's incredible potential for…
Read
Childhood And Education #19: Letting Kids Be Kids #2
via Substack Zvi [999] — I cannot emphasize enough the need to let kids be kids.
Read
Google Search is getting its biggest changes ever
via The Verge AI [4] — Google Search is entering the next phase of its AI evolution. During Google I/O 2026, the company showed off a reimagined search box that makes it easier to flow between AI Overviews, the AI-generated summaries that appear at the top of search results, and…
Read
AgentWall: A Runtime Safety Layer for Local AI Agents
via ArXiv cs.AI [8] — The safety of autonomous AI agents is increasingly recognized as a critical open problem. As agents transition from passive text generators to active actors capable of executing shell commands, modifying files, calling APIs, and browsing the web, the…
Read
Thoughts on interviewing candidates for AI safety fellowships
via LessWrong AI [5] — Around July last year I decided I was going to go all in on technical AI safety research. To do that I’d need to get into an AI safety fellowship, quit my job, and sell everything that was in my flat in South Africa (hopefully in that order).I applied to…
Read
Classifier Context Rot: Monitor Performance Degrades with Context Length
via LessWrong AI [3] — Monitoring coding agents for dangerous behavior using language models requires classifying transcripts that often exceed 500K tokens, but prior agent monitoring benchmarks rarely contain transcripts longer than 100K tokens.We show that when used as…
Read
Dating Roundup #12: Sex and Violence
via Substack Zvi [999] — No more burying the sex stuff under an avalanche of other stuff so no one notices.
Read
Fast-tracking genetic leads to reverse cellular aging
via DeepMind Blog [4] — Biologists use Co-Scientist to find novel factors that successfully rejuvenate human cells.
Read
Verifiable Agentic Infrastructure: Proof-Derived Authorization for Sovereign AI Systems
via ArXiv cs.AI [3] — Modern cloud and enterprise systems rely on identity-centric authorization, assuming that callers possessing valid credentials are safe to execute commands. The emergence of autonomous AI agents invalidates this assumption: agents can generate syntactically…
Read
SDOF: Taming the Alignment Tax in Multi-Agent Orchestration with State-Constrained Dispatch
via ArXiv cs.AI [3] — Multi-agent orchestration frameworks such as LangChain, LangGraph, and CrewAI route tasks through graph-based pipelines but do not enforce the stage constraints that govern real business processes. We present SDOF, a framework that treats multi-agent…
Read
An Introduction to Exemplar Partitioning for Mechanistic Interpretability
via LessWrong AI [7] — Most of what we currently call "feature discovery" in language models is wrapped up in dictionary-learning methods like sparse autoencoders (SAEs) – which work, and which have been scaled to millions of features on frontier-scale models, but which bundle…
Read
A Year Late, Claude Finally Beats Pokémon
via LessWrong AI [3] — Credit: ClaudePlaysPokemon Elevator Shanty by KurukkooDisclaimer: like some previous posts in this series, this was not primarily written by me, but by a friend. I did substantial editing, however.ClaudePlaysPokemon feat. Opus 4.7 has finally beaten…
Read
A Two-Dimensional Framework for AI Agent Design Patterns: Cognitive Function and Execution Topology
via ArXiv cs.AI [3] — Existing frameworks for LLM-based agent architectures describe systems from a single perspective: industry guides (Anthropic, Google, LangChain) focus on execution topology -- how data flows -- while cognitive science surveys focus on cognitive function --…
Read
The hard core of alignment (is robustifying RL)
via LessWrong AI [5] — Most technical AI safety work that I read seems to miss the mark, failing to make any progress on the hard part of the problem. I think this is a common sentiment, but there's less agreement about what exactly the hard part is? Characterizing this more…
Read
Risk reports need to address deployment-time spread of misalignment
via Alignment Forum [999] — Risk reports commonly use pre-deployment alignment assessments to measure misalignment risk from an internally deployed AI. However, an AI that genuinely starts out with largely benign motivations can develop widespread dangerous motivations during…
Read
Mechanistic estimation for expectations of random products
via Alignment Forum [999] — We have developed some relatively general methods for mechanistic estimation competitive with sampling by studying problems that are expressible as expectations of random products. This includes several different estimation problems, such as random…
Read
Monthly Roundup #42: May 2026
via Substack Zvi [999] — At least we probably won’t have another pandemic.
Read
No headlines on this page match those filters.
Reference points
Hinton, Bengio, and other signatories call for global attention to AI extinction risk.
CAIS statement, 2023 →
OpenAI describes agents bypassing isolation and compromising third-party systems during internal evaluations.
OpenAI incident disclosure, August 2026 →
Yudkowsky argues that further frontier development poses unacceptable danger.
Yudkowsky's essay, March 2023 →