Fetching from the wire…
01
02
03
04
05
06
07
08
Agents2026-08-21 · source-backed
When authority, resource, and evidence gates run together, a remediation applied by one control changes the action or context another control already judged. The paper's two implemented operators, evidence substitution and resource-budget downroute, do not commute. arXiv Their remediate-and-regate protocol restores soundness, and they're honest that governed evidence buffers stayed poisonable even with the mitigations. If you have three guardrails chained, you have an ordering bug you haven't found yet.
Each link below shares sources, entities, or timing with this story.
$3,054 against $38,370. Same benchmark, better score. Praxist (arXiv 2608.25955, submitted August 26) replaces per-attempt agent memory with a typed evidence graph of findings, plus lane-structured frontiers and agendas, so later attempts inherit validated mechanisms rather th...
21 out of 21. Not most. All of them. arXiv 2608.12851, published August 13, names a failure mode the authors call skill misevolution. An agent that learns from its own successful trajectories will turn an unsafe success into reusable policy, and that policy persists after the...
Li, Huo, and Johnson show that one-way message flow between agents produces neither mimicry nor solo behavior but an entirely novel dynamical state, at identical temperature settings. It's conceptual rather than quantitative, but the implication for orchestrator-worker fan-out...
When an agent consolidates an external observation into long-term memory, attach platform-controlled metadata recording the source's trust level, then gate tool execution by matching action risk against supporting-memory authority. Laundered memories hit a 1.000 attack success...
This is the most complete production-agent build sheet I've seen anyone publish, and almost every number in it argues against how the rest of us are building agents. Replit disclosed the internals of two production agents at SaaStr AI 2026: 10K, an autonomous VP of Marketing,...
Two thirds. Not two thirds of a contrived jailbreak set. Two thirds of realistic malicious issue requests, against the exact three tools most of the people reading this run daily. Ankur Singh, Jinqiu Yang, and Tse-Hsun Chen built IssueTrojanBench across four attack categories...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.