Topic
Everything filed under AI, newest first.
RSS · JSON · All topics
What changed in how we evaluate AI SRE agents
Why a coding harness is not an AI SRE agent, and the expanded criteria One2N now uses to evaluate AI SRE agents: triggers, context, RCA quality, audit and guardrails.
15 min · 3,505 words
Making Kubernetes AI-Ready for NVIDIA Vera Rubin on Day One
Configuring and optimizing Kubernetes for new GPU hardware such as NVIDIA Vera Rubin is complex and error-prone. Drivers, kernels, runtimes, networking, operators, and schedulers must work together, with settings tuned to the hardware and workload. Small mismatches can prevent workloads from running or limit performance. The NVIDIA AI Cluster Runtime (AICR)—part of the NVIDIA DSX AI Factory Platform—simplifies this work by capturing tuned configurations in reusable recipes, generating bundles for existing deployment tools, and validating the running cluster. It will help…
12 min · 2,706 words
Nace.AI releases Drex for Finance, a decision model for fintech fraud, credit and filings that returns option probabilities instead of free text, scoring 77.3 on its finance index versus 63.6 for Jev 1.14.
5 min · 1,241 words
Parse Long PDFs Without Breaking Tables Using Unlimited-OCR-2
OpenTeams Engineering explains Unlimited-OCR-2, which reads many PDF pages in one request so tables split across pages stay intact for downstream parsing.
6 min · 1,291 words
Planning an AI-assisted job search in 2026
Lyndsy Simon documents how he is using AI agents and automation to structure a sudden job search—resume, outreach, and tracking—after years of relying mainly on his professional network.
4 min · 971 words
Nyquist Limit: the frequency of changes in tech is invalidating our understanding
Ryan Prior argues that LLM-driven increases in the rate of software change outpace how fast humans can observe and understand systems, so teams end up measuring noise instead of signal, borrowing the Nyquist limit from signal processing.
3 min · 744 words
Progress on provable inference for large language models
Anthropic's Frontier Red Team and collaborators from EPFL, MIT and elsewhere describe progress on cryptographic proofs that an LLM's output token stream really came from a specified model, including a prototype system, its performance costs, and open problems for verifiable inference.
21 min · 4,805 words
Klang Research sent the same 3,000 sentences through twelve frontier AI models and found every one charges 30% or more extra tokens for Swedish, Norwegian, Danish and Finnish than for English, explaining where the extra tokens come from.
5 min · 1,037 words
The shelf life of LLM findings
Stanford's Owen Queen and James Zou re-ran 50 widely cited findings about language models, from sycophancy and jailbreaks to fabricated citations and long-context limits, across 20 models released July 2024 to September 2026, finding that within 15 months most claims had expired or substantially weakened.
28 min · 6,536 words
The Rise of LLMJacking: How Threat Actors Monetize Compromised Cloud and Model Resources
Google Threat Intelligence Group and Mandiant describe 2026 incidents where attackers hijacked enterprise LLM access across cloud providers, running up fraudulent charges above $100,000 in a day and up to $1 million cumulatively, plus how the access is resold, and detection, hunting and mitigation guidance with IOCs.
13 min · 3,080 words
so i guess everyone uses llms to think now
Security researcher eva explains why she never uses AI for the research on her blog: tinkering is the fun, she can't take credit for an LLM's work, and she argues AI firms use hyped vulnerability disclosures, PoC videos and hash drops (php2shell, 'tailworm') as marketing.
8 min · 1,760 words
Mathematics has changed. Let’s act like it
A guest post by mathematician Frank Calegari on Terence Tao's blog arguing that after OpenAI's release of 700+ AI-written research papers, mathematicians should stop mourning and instead change how they hire, publish, referee and train students for a world where AI can make real contributions.
17 min · 3,877 words
Who really runs the group project? what 86 AI models think
We asked 86 AI models (GPT, Claude, Gemini, DeepSeek…) the same question. 77% picked The one who does all the work. See every answer and who dissented.
15 min · 3,483 words
Energy-Based Reasoners: Your Recurrent Reasoner Should Be an Energy-Based Model
Researchers from Duke, UIUC and Harvard introduce Energy-Based Reasoners, looped reasoning models that learn an energy function over latent states; with 1.8M parameters they reach 99.9% on Sudoku-Extreme using 345x less inference compute than the next-best model.
4 min · 873 words
Why is Gemini 4 Carbon taking so long?
Kernel traced every step of Gemini 4 Carbon's runs on a private eval to explain why it spent close to an hour on tasks that Opus and Astra finished in minutes, pointing to behavioral issues and mechanical inefficiencies such as fewer tool calls per step and frequent re-reading.
5 min · 1,063 words
Could a modern chess engine on a Pentium have defeated Kasparov in 1996?
Nicholas Carlini ports Stockfish to a 1995 200 MHz Pentium Pro with 64MB RAM, estimates it plays around 3300 Elo at classical time controls, and argues that algorithmic progress alone would have let commodity 1996 hardware beat Kasparov without Deep Blue, with lessons for software versus hardware gains in AI.
14 min · 3,118 words
Kunal (knlb.dev) argues that uncritically accepting whatever coding agents produce is fast and seductive but erodes understanding, and lays out practical defenses: choosing a 'vibe limit', watching for automation bias, forcing yourself into the loop, demanding comprehensible output, delegating deliberately, and leaning on good design.
12 min · 2,731 words
It wasn't AI that flipped Proton's charge
Longtime Proton customer Raed Chammam argues the backlash over Proton's AI push is really about leadership: a CEO calling an engineer's request for pentesting time "loser thinking" undermines the careful engineering a privacy company sells, regardless of whether code is AI-written.
2 min · 456 words
Andrej Karpathy describes building HackerTube, a YouTube-style site made only from videos recommended on Hacker News, with Opus 5.5 in about two hours, with local-only recommendations and a data export.
1 min · 248 words
Is it a red flag if they don't have Instagram? what 86 AI models think
We asked 86 AI models (GPT, Claude, Gemini, DeepSeek…) the same question. 99% picked No. See every answer, who searched the web, and who dissented.
10 min · 2,294 words