Fetching from the wire…
01
02
03
04
05
06
07
08
Security2026-08-30 · source-backed
A 170-artifact, 145-family benchmark scored three scanners on whether they return a decision at all. ModelAudit produced a definitive verdict on 100% of the 135 labeled families, Fickling on 81.5%, ModelScan on 49.6%. But ModelScan reached 100% precision, recall and F1 on the subset it judged. (arXiv 2608.27424) Treat an N/A as unscanned rather than clean, and read a scanner's coverage rate before you trust its published F1 on a supply-chain gate.
Each link below shares sources, entities, or timing with this story.
arXiv 2607.24392 measured secondary costs across downstream task performance, over-refusal on benign inputs, and inference cost. Rule-based defenses best preserve task performance. Conservative self-reflective defenses drive the most over-refusal. Multi-round defenses carry th...
Beyond binary human-vs-machine detection: identifies which specific LLM generated a code snippet. Enables vulnerability triage (which model produced the bug?), licensing audits, and distillation detection. Directly relevant to the Anthropic distillation crackdown. (arXiv 2603....
Somebody finally measured the thing every one of us suspected. Researchers analyzed 247,694 instruction lifetimes across 1,867 repositories and found agentic instruction files grow without bound, more than tripling over their lifetime at +4.9 net instructions per commit, with...
arXiv 2608.09867 finds encrypted chain-of-thought blocks move across sessions, users, and models, and carried recoverable PII and credentials into public repositories. Providers pushed reasoning state to the client to avoid storing it server-side, which quietly made it your li...
The Replay Gap (arXiv 2608.08239) forked live SWE-bench trajectories at controlled points, rebuilt the environment, and continued each fork with a different model across ~900 rollouts. Model swaps rewrote 61–94% of post-fork actions and diverged at the very first post-fork act...
arXiv 2608.03626 restructures the lifecycle around security boundaries rather than workflow efficiency: 32 stages across Data, Model, Distribution and Application layers plus a 12-stage LLMOps pillar and 9-category governance pillar, with 13 stages newly separated because they...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.