Fetching from the wire…
01
02
03
04
05
06
07
08
Policy2026-09-18 · source-backed
His September 16 essay argues AI vulnerability research solves the wrong problem, since finding vulnerabilities was never the bottleneck, getting packages updated is. He says he now spends upwards of 75% of his time dealing with AI directly or indirectly, and that the engineering hours poured into AI bug discovery could have gone to asset inventory and automated patching. His second point is harder to dismiss: AI-generated patches are increasingly reviewed by AI, leaving the humans on call with steadily less comprehension of systems getting more opaque.
Each link below shares sources, entities, or timing with this story.
Willison quoted Florian Herrengt on August 12 about teams accumulating so many AI-generated layers that nobody retains a working model of the system. The artifact still works, the org loses the mid-level engineers who could reason about why. Hold that against the OpenAI paper...
Introduces temporal causal diagnostics to distinguish legitimate task execution from injected manipulation in multi-turn agent interactions, plus context purification to neutralize poisoned content. Directly applicable to anyone building agents that call external tools. arXiv...
Google calls it the biggest Search change in over 25 years, rolled out through mid-June. Search answers the query directly and builds a page around the answer rather than returning a list. For anyone shipping content, this is a structural hit to click-through economics. The an...
A 15-run pilot, a pre-registered 20-run confirmatory ablation and a pre-registered 2x2 factorial with 40 runs across two vulnerable lab systems (arXiv 2609.15887). Removing verification raised reported findings (median 2 against 0, p = 0.00003) and cut precision (0.353 against...
Ryan Lopopolo's essay argues that outside your own expertise you have nothing to check a model against except its priors, and those priors were shaped by non-experts rewarding output experts would call poor. Since there's no unhackable grader and models are rewarded for effici...
The complete public record preserves not just what each agent wrote but what it could see before writing, and one rule governs all three arrival decisions (where to write, what to call itself, how to word the message): an agent picks an option with probability close to that op...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.