Blog — Page 37
Engineering notes, case studies, and lessons from the field.
NVIDIA Open-Sources Ising — The First AI Models Built Specifically for Quantum Computing
NVIDIA's Ising is an open-weight family of AI models for quantum processor calibration and error correction decoding. The 35B calibration model outperforms GPT-5.4 on QCalEval, and the decoding model runs 2.5x faster with 3x better accuracy than classical methods.
Claude Opus 4.7 Is Out — 87.6% on SWE-bench, New Agentic Controls, Same Price
Anthropic's latest flagship model scores 87.6% on SWE-bench Verified and introduces a new xhigh effort tier and token budget system for agentic loops — all at the same $5/$25 per million tokens as Opus 4.6.
CVE-2026-3854: GitHub Enterprise Server RCE — 88% of Self-Hosted Instances Still Unpatched
A CVSS 8.7 flaw in GitHub Enterprise Server lets any authenticated user execute arbitrary commands with a single crafted git push. Patches have been available since March, but 88% of self-hosted instances were still exposed at disclosure.
Big Tech Earnings Sweep: Google Cloud +63%, Azure +40%, AWS +28% — AI Spending Is Paying Off
Alphabet, Microsoft, Amazon, and Meta all beat Q1 2026 estimates on April 29. Google Cloud grew 63% to $20B with a $460B backlog, and Alphabet's net income jumped 81% — the clearest signal yet that AI infrastructure investment is generating real returns.
OpenAI Lands on Amazon Bedrock the Day After Microsoft's 7-Year Exclusivity Ends
GPT-5.5, Codex, and Managed Agents are now in limited preview on Amazon Bedrock, giving AWS customers native access to OpenAI's frontier models for the first time. The move dismantles the Azure lock-in that defined the AI cloud market since 2019.
CVE-2026-42208: Pre-Auth SQL Injection in LiteLLM Actively Exploited — AI API Keys at Risk
A CVSS 9.3 SQL injection in LiteLLM's authentication path was weaponized within 36 hours of disclosure. Unauthenticated attackers can extract OpenAI organization keys, Anthropic console credentials, and AWS Bedrock IAM tokens without any login.
NVIDIA Nemotron 3 Nano Omni Is Open, Multimodal, and 9x Faster Than Its Competition
NVIDIA has open-sourced a 30B-parameter multimodal model that processes text, vision, audio, and video in a single inference pass. It outpaces rival omni models 9x on throughput while topping six industry leaderboards.
Paragon Is Stonewalling Italian Prosecutors Investigating Graphite Spyware Attacks on Journalists
More than a year after Italian prosecutors opened criminal investigations into Graphite spyware attacks on journalists, Israeli-American vendor Paragon Solutions has not responded to formal legal assistance requests routed through the Israeli government — raising questions about accountability for commercial spyware firms.
CVE-2026-25874: Critical Unpatched RCE in Hugging Face's LeRobot Puts Robotics Hardware at Risk
A CVSS 9.8 vulnerability in LeRobot's PolicyServer uses Python pickle deserialization over an unauthenticated gRPC channel, giving remote attackers full code execution on GPU inference servers and the connected robot clients they control — no patch available yet.
Page 37 of 61 · 549 articles