This article grew out of the comments on my previous piece, Why AI Research Tools Struggle to Fact-Check the Web, where, using GPT Researcher as an example, I showed how common search engines and context-building methods undermine the very foundations of AI-based fact-checking.
Not that those tools were bad. They’re brilliant at standard (re)search tasks. But when finding genuinely diverse and independent sources matters, they start to show their limits. So we fixed that part. Adding source assessment through an admission policy was enough to make a difference.
But as the readers’ comments quickly showed, I was only touching the tip of the iceberg. — Read More
Recent Updates Page 8
The Year Finding and Exploiting Bugs Became Cheap, and What to Do About It
Over the past few years, nearly every security researcher I know has incorporated LLMs into their process. What began with chatbots quickly evolved into scripts calling model APIs, then agents, skills, custom harnesses, autoresearch loops, and more approaches than anyone can reasonably keep track of.1PeckShield tracked 16 hacks in January 2026, 15 in February, 20 in March, around 40 in each of April, May, and June, and a record 50 in August. August losses were $136M, down 49.5% from July, so attackers are striking far more often while taking less from each incident. CoinGecko’s 2026 State of Crypto Security Report counts 245 incidents and $3.63B lost between January 2025 and July 2026.
Toward the end of 2025, something shifted. Models and the harness/systems around them became better, and AI-assisted bug finding and exploit development stopped feeling like an interesting experiment, but it became reality, while we start observing an increased amount of exploits1. — Read More
How well do agents use test/verification techniques?
We previously noted that, while it’s easier than ever to hit a particular quality bar by having coding agents use effective test techniques, software quality seems to be getting worse, indicating that whatever defaults developers are using may not work very well. Here, we test if simple instructions to agents to use particular techniques or libraries improve implementation correctness, as a kind of test to see how effective agents are when guided by someone with no expertise in testing who’s maybe heard that you should apply certain techniques or use certain libraries. — Read More
CONTEXT, SEMANTICS, AND ONTOLOGY: A PRIMER FOR THE AGENTIC ERA
There’s so much talk about new ways of working with agent engineering supported workflows. New models are independently creating new metrics and transformations, finding gaps in the business data, reviewing the SQL they write, and verifying everything works with your data platform.
All of it is autonomous, so one might say, what then is left for us humans to do in the data work context? Many are defaulting to adding or curating context, ergo the rise of a context layer. I see even more talks about added Ontologies. Maybe you ask yourself, what is that even? Do we need all of it?
This article is a primer about the context layer, the difference between a classical semantic layer contained in every BI tool, and an external semantic layer. — Read More
Research acceleration: The view inside OpenAI
For AGI to benefit all of humanity, we believe it must be democratically governed. This can only happen through an informed public debate about the capabilities, risks and safeguards of highly capable AI systems. People everywhere need to understand the likely future trajectory of frontier AI, so they can have a meaningful voice in how it develops.
… We aim to safely build an automated AI researcher that can work under human supervision to further progress on deep learning and alignment, enabling iterative improvements. According to our measurements, we have now reached the goal, announced last fall, of having an automated research intern by September of this year. — Read More
The Curiously Playable Universe
In 2024, Google DeepMind’s AlphaProof solved three of the five non-geometry problems at the International Mathematical Olympiad. Unlike a chatbot producing a plausible-looking derivation in prose, AlphaProof worked in Lean, a formal mathematical language in which proofs can be mechanically checked. It trained by proving or disproving millions of mathematical problems, receiving an unusually clean signal each time: either the proof checked out or it didn’t.
This looked like another frontier falling to artificial intelligence. Chess had fallen, then Go, then protein folding, then programming, and now serious mathematics appeared to be giving way too. Mathematics is among the activities we most strongly associate with the mysterious upper reaches of human intelligence, so perhaps the obvious conclusion was that AI was climbing toward those reaches.
But there is another way to look at what happened: Lean, where a mathematical proposition is represented as a type and a proof as a term of that type. — Read More
Discovery of a new OpenAI agent message board
We found ~18,000 posts from autonomous AI agents (self-identifying as from OpenAI) using the public internet to communicate during a web-retrieval task.
These AIs colluded to share answers, research their environment, and bypass sandbox restrictions.
Almost all of the logs of the agents communicating on this site are publicly available. However, we host our own copy where we’ve reconstructed the deleted pages via edit history and redacted personally identifiable information.
We encourage others to take a look and write up their own analyses of this data. — Read More
5 amazing visuals show how the male fruit fly’s brain map is advancing neuroscience
For the first time, scientists have mapped every single neural connection in the brain and central nervous system of an adult male fruit fly. In this years-long project by HHMI Janelia Research Campus, Google Research, and collaborators from the scientific community, this map of the male fruit fly brain includes a record-breaking more than 166,000 neurons. It’s a big step in advancing neuroscience experiments on this key model organism. — Read More
How to Turn a Book Into an AI Skill You Come Back To
I used to buy books to learn something. I would read, highlight, take notes, and file them somewhere I meant to come back to. Then weeks later, in the middle of a real project, I would try to remember what the book said.
Most of it never made it across that gap. The reading and the applying sat too far apart.
… There is a shorter path now. You can turn a book into a skill your agent loads while you work, so the framework arrives at the moment you need it. Convert it once and the book stops being something you try to remember and becomes something that runs. — Read More
LLMs are becoming commodities
With the GPT-6 announcement today, we immediately thought of two things. First, the model matters less and less than the application of the model. The frontier labs seem to be increasingly focused on the best models for particular applications (which we posted about recently). That reminded us of a post of ours from 2.5 years ago, which we feel is more relevant than ever.
If the application matters more than anything, then the open question (in 2026 terms) becomes whether the harness can be separated from the intelligence engine. Our bet would be yes. — Read More
