Sophos and OpenAI report that AI agents now resolve 52% of Sophos MDR cases end to end and cut agent-handled investigation time from about 38 minutes to about 89 seconds. The numbers are real as reported, but they were calculated by the companies selling the arrangement. Here is what they measure, what they do not, and what customers control.
LegalOn Technologies says it cut estimated daily Codex costs by roughly 65 percent using model tiering, restricted Fast mode, and layered budget caps. The figures are vendor-reported and unaudited, but the governance structure behind them is documented in detail, and it is the part most organizations have not built.
NASA's SSPICY mission, built and operated by Starfish Space, launched Oct. 1, 2026, to demonstrate autonomous inspection of inoperable satellites in low Earth orbit. Here is what it tests, what it does not, and why inspection is a prerequisite for satellite repair and debris removal.
NASA announced on October 8, 2026 that L3Harris Technologies will build the Engineering Test Unit of the all-glass telescope for the LISA gravitational-wave observatory, the final pre-flight step before flight hardware for a mission that will listen to the universe in frequencies no Earth-based detector can reach.
NVIDIA's October 7, 2026 technical blog describes DSX Air, a node-based digital twin that lets AI agents rehearse and validate AI factory changes in simulation before hardware arrives, with a human approval gate before production. All results are vendor-described.
NVIDIA's new open-source CUDA-X toolkit keeps scientific image data on the GPU from sensor read through classification. On representative multi-GPU workloads, NVIDIA reports up to 14,900x image-loading and 14,550x signal-processing speedups versus an x86 CPU baseline. Here is what that does and does not mean, and why the Rubin Observatory's 60-second alert window shows why it matters.
NVIDIA's cuOpt team released mPDLP, a solver that spreads massive linear programming problems across NVLink-connected GPUs, with vendor-reported results suggesting what planners could do with models that once took overnight.
A new benchmark from two Berkeley researchers grades AI models on schematic transcription and device sizing using 273 topologically distinct open-source circuits from Tiny Tapeout manufacturing shuttles. The results reveal sharp limits in visual reasoning, and a refusal problem that has nothing to do with skill.
CUDA 13.1 brings green contexts to the mainstream Runtime API, letting an application reserve a fixed slice of streaming multiprocessors so a small, latency-critical job never waits behind a bulk workload. NVIDIA reports 0.007 ms versus 0.140 ms on a Blackwell GPU; all figures are vendor-reported.
Google DeepMind released EmbeddingGemma 2, a 740-million-parameter open model under Apache 2.0 that maps text, images, video and audio into one shared embedding space, enabling private offline media search on consumer hardware.
TII's new dialect-specialized model scored 0.52 on LLM-judged dialect fidelity in a vendor-run evaluation, against 0.05 or lower for competing models that mostly replied in Modern Standard Arabic. The result, if it replicates, says something important about how AI handles the languages people actually speak.
A new preprint tests whether frontier AI coding agents can edit working games like Minecraft and Terraria without breaking them. The best configuration solves 78.2 percent of tasks, but most failed edits still build and load. The lesson for software teams: a passing build is not proof an AI edit did the right thing.
A new arXiv preprint classified 9,821 annual reports from 1,362 UK listed companies and found that AI risk mentions jumped from 2.8% to 41.2% in five years, while only 4.3% of 2025 reports contain disclosure the study deems substantive. The gap between boilerplate and substance is the story.
A joint NVIDIA and Nscale evaluation reports that policy-governed power sharing let 192 GPUs run under a 264.4 kW budget sized for 140, with 49.2 percent more aggregate throughput and a measured cost in tail latency. All numbers are self-reported vendor measurements.
The open-source runtime enforces permissions outside the agent workload, keeps credentials out of reach, and uses formal logic to prove what a policy allows. Here is what is demonstrated, what is claimed, and what remains unproven.
New machine-readable skill files give AI coding agents verified DOCA API signatures and hardware constraints for BlueField DPUs. NVIDIA says agents without them met only 19 percent of graded checklist items; with them, 100 percent. The vendor-run evaluation is the story, and so is its caveat.
OpenAI's October 2 documentation for the GPT-6 family tells builders to cache aggressively, right-size reasoning effort, and set human decision boundaries. It is useful, documented guidance, and it is also advice from the party selling the tokens.
A new arXiv paper describes a free, MIT-licensed AI research assistant that runs on the scientist's own computer, keeps a lab notebook whose entries can never be erased, and records what the AI actually did in a log the AI cannot write to. The benchmark result is the authors' self-reported evaluation, and one design limitation sits in the abstract, not a footnote.