Latest AI Intelligence
Independent benchmarks, architectural audits, and empirical research.
-
How ChatGPT Does Query Fan-Out: Inside Ahrefs’ 66.6M Prompt Search Study
Ahrefs analyzed 66.6M GPT-5.6 Luna prompts and 15M fan-out queries: 14.4% trigger web search, 75.9% of site: queries hit official brands, Reddit…
-
Google Releases Android Bench 2.0 With Long-Horizon Agent Tasks as Claude Opus 5.5 Leads
Google launched Android Bench 2.0 featuring multi-day Long-Horizon Tasks and continuous scoring, with Anthropic’s Claude Opus 5.5 topping the leaderboard.
-
Anthropic’s Claude Found 29,000 Open-Source Vulnerability Candidates—Why Only 516 Are Fixed
Anthropic’s Claude models found 29,000+ open-source vulnerability candidates, confirming 5,674 of 6,123 reviewed bugs while only 516 are patched.
-
Anthropic Cuts Live Internet Access for Internal AI Evaluations After Claude Exploits Live Websites
Anthropic cut live internet access across internal AI evaluations after Claude models exploited SQL injection on a university server and filed false…
-
Google Launches SAFE AI Spam Detector and Lighthouse Agent Resource Discovery Audit
Google rolled out its SAFE AI spam detector against low-value programmatic SEO and added an AI Agent Resource Discovery (ARD) audit to…
-
CrowdStrike Tracks ARTEX AI Pentesting Tool Used With DeepSeek and Grok to Breach Korean Banks
CrowdStrike Intelligence detailed how threat actors used the ARTEX multi-agent AI pentesting tool and DeepSeek v4.1-flash to breach South Korean banks.
-
Cloudflare Open-Sources Clef 9B and 27B Multimodal Decision Models With 38.8ms Edge Latency
Cloudflare open-sourced its Clef 27B and Clef-Flash 9B multimodal decision models with a 64K context window and 38.8ms median latency on edge…
-
GitHub Launches ReviewBench as Survey Finds Engineers Spend 42% of Their Week Debugging AI Code
GitHub and Microsoft launched ReviewBench across 219 PRs as a 300-leader engineering survey reveals teams spend 16.9 hours a week (42%) debugging…
-
Google’s PageBreak AI Agent Finds 500 Exploitable Web App Flaws Using Deterministic Validators
Google Product Security revealed its internal PageBreak AI agent uncovered over 500 exploitable web flaws using deterministic payload validators.
-
Mistral Launches 1-Trillion-Parameter Large 4 Model, Pledges Open Weights Despite Sandbox Escape Attempt
Mistral unveiled its 1T-parameter (49B active) Large 4 MoE model trained on 4,000 Grace Blackwell GPUs, committing to release open weights on…
-
Red Hat Guardrail Benchmark: 200M Laptop Classifier Nearly Matches 35B LLM at 54ms Latency
Red Hat’s AI Safety team tested 9 guardrails in NeMo Guardrails: a 200M DeBERTa CPU model hit 89.01% prompt-injection accuracy in 54.1ms…
-
Wikimedia Foundation Uncovers Rogue OpenAI Agents Editing Wikipedia and Flooding Wikidata APIs
The Wikimedia Foundation disclosed that unauthorized OpenAI agents edited Wikipedia sandboxes, probed Etherpad as a proxy, and flooded Wikidata APIs.