Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-08-25 · high
Math sped up too, with arXiv submissions doubling in some subfields, plus new results on the Jacobian conjecture and Green's Problem 44.
Why now: The comparison, 2026 against 2025, appeared in Jack Clark's Import AI newsletter on August 24.
Vulnerability discovery sped up in 2026 across cURL, OpenSSL, Firefox, Microsoft, the US NVD and OSV. That comparison against 2025 comes from a METR analysis in the August 24 issue of Import AI.
Security teams evaluating AI-assisted vulnerability triage now have a real 2026 data point to weigh. Anyone counting on the same models to out-optimize hand-tuned systems doesn't.
The same analysis found zero measurable acceleration across seven optimization benchmarks, a different kind of task than fixing a bug in existing code. They cover CIFAR-10, Hutter compression, Gurobi MIP, MIPLIB, nanoGPT, Stockfish and the matrix-multiplication exponent.
Math moved too. arXiv submissions doubled in some subfields, and the analysis credits AI-assisted work toward the Jacobian conjecture and Green's Problem 44.
Jack Clark, who covered the analysis, reads the split as per-skill phase changes, not one smooth industry-wide lift. Coding jumped in 2025. Cybersecurity jumped in 2026. Optimization hasn't had its turn.
Each link below shares sources, entities, or timing with this story.
Import AI 466 pairs two results: Anthropic showed Opus 4.7 completing a robot task in 9 minutes against 181 minutes with earlier technology, and Sunday Robotics' ACT-2 hit 99.1% ±0.3% success across 785 autonomous laundry-folding attempts spanning 9 garment types in homes it h...
The Fields medalist published his assessment August 12, arguing models excel exactly at problems amenable to "try lots of things till you get lucky" and not at the search-tree pruning humans use for proof discovery (Gowers). His observation about the recent run of ten solved m...
The RSI debate has been vibes and timelines for two years. This week a frontier lab published an actual measurement from inside its own walls. The Anthropic Institute reported an 8x increase in lines of code merged into its codebase in 2026 versus the 2021–2024 baseline. The t...
The Model Context Protocol has a security problem that's no longer theoretical — it's statistical. Between January and February 2026, researchers filed 30+ CVEs against MCP servers, clients, and infrastructure. One package with nearly 500,000 downloads carried a CVSS 9.6 RCE....
The July 30 changelog closed the hosted model-catalog and playground service that let developers prototype against multiple LLMs from GitHub directly. If you prototyped against Models endpoints, this is a migration event, not a skim. The surrounding changelog items (Copilot up...
On June 16–17, Google extended its A2A interoperability push with a standard for how agents discover available resources and tools, the same week Microsoft's Work IQ went GA with A2A plus remote MCP. It's a multi-vendor race to standardize the plumbing. Design against the inte...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.