MICROSOFT
30 days · UTC
Synchronizing with global intelligence nodes...
Anthropic launches Project Glasswing, giving controlled access to Claude Mythos for vulnerability discovery
Anthropic formed Project Glasswing and is withholding its Claude Mythos Preview model for controlled, defensive use after it found thousands of high‑s...
Oracle-SWE dissects the “oracle hints” behind SWE-bench wins, challenging headline coding benchmarks
New research isolates which “oracle” hints actually move SWE-bench agent scores, explaining why headline results often don’t match real coding impact....
Anthropic previews Claude Mythos and launches Project Glasswing to weaponize defense against zero‑days
Anthropic previewed Claude Mythos and launched Project Glasswing, claiming the model can autonomously find high‑severity bugs across major OSes and br...
VS Code 1.115 previews multi-repo Agents app; Microsoft open-sources Agent Governance Toolkit for OWASP risks
Microsoft previewed a VS Code Agents app in VS Code 1.115 and released an open-source Agent Governance Toolkit aligned to OWASP to secure agent workfl...
Anthropic’s Mythos and Project Glasswing push AI into real-world vuln discovery, with tight access and strong benchmark signals
Anthropic launched Project Glasswing and a Mythos Preview model that finds serious software bugs, pairing industry partners with restricted access and...
Anthropic launches Project Glasswing and restricts Claude Mythos Preview to harden critical software
Anthropic launched Project Glasswing and a restricted Claude Mythos Preview, a model that reportedly finds thousands of serious software vulnerabiliti...
Microsoft ships in-house MAI models for speech, voice, and images, aiming for lower GPU cost and enterprise scale
Microsoft launched three in-house MAI models for transcription, voice, and images, targeting better accuracy, speed, and cost than current options.
GitHub flips Copilot training to opt-out on April 24; Copilot CLI 1.0.13 brings MCP inference approvals, rewind, and speedups
GitHub will start training Copilot on user interaction data by default on April 24 while Copilot CLI ships notable agent/MCP improvements. GitHub pla...
GitHub flips Copilot training default for individuals: interaction data opt-out starts Apr 24
GitHub will start training Copilot on individual users’ interaction data by default on April 24 unless they opt out; Business and Enterprise accounts ...
Coding agents in production: architecture choices, reliability budgets, and hitting the brakes
A wave of practitioner write-ups agrees: shipping coding agents is about reliability budgets and the right architecture, not flashy demos. At the AAA...
From copilots to coworkers: the agentic turn hits the enterprise
Enterprise AI is shifting from chat assistants to autonomous agents that act on your behalf. Microsoft is reportedly rebranding Copilot as “Friday,” ...
Malicious Windsurf IDE extension steals dev credentials via Solana; tighten IDE extension hygiene
A fake Windsurf R-language extension used the Solana blockchain to drop a NodeJS stealer that targets developer credentials. Bitdefender found a Wind...
Cursor 2.6.19 regressions reinforce that enterprise AI coding is a control‑plane decision
Fresh Cursor 2.6.19 regressions underscore that enterprise AI coding choices hinge on governance, failure modes, and rollback. A comparative [write-u...
MCP grows up: Chrome DevTools control, C# SDK 1.0, and early WebMCP
MCP tooling is rapidly maturing with a C# SDK 1.0, a Chrome DevTools MCP server for reliable browser automation, and early WebMCP experiments for agen...
OpenAI vs GitHub: enterprise push and rising lock‑in risk
OpenAI’s enterprise push and a reported GitHub rival raise new lock-in and architecture questions for teams adopting AI across the SDLC. OpenAI is re...
From Basic RAG to Agentic and GraphRAG: A Production Blueprint
A practical series shows how to evolve basic RAG into agentic, adaptive, and graph-backed systems that cut cost and raise answer quality for real prod...
Copilot CLI 0.0.422 lands automation-friendly upgrades as VS Code previews agent plugins
GitHub shipped Copilot CLI 0.0.422 and VS Code previewed agent plugins, tightening how AI agents run across terminal, editor, and CI workflows. Copil...
Meta locks down news training data and centralizes AI delivery as OpenAI eyes a GitHub rival
Meta is formalizing AI training data access and centralizing AI deployment while OpenAI reportedly builds a GitHub rival, signaling a consolidation of...
Amazon Q vs GitHub Copilot in VS Code: Speed vs Rigor
In a head-to-head VS Code test of agentic AI for a complex editorial workflow, Amazon Q Developer completed the task faster with less rework, while Gi...
Copilot CLI locks down MCP; Skills mature; watch VS Code and licensing gotchas
GitHub Copilot’s latest CLI releases tighten Model Context Protocol access and add workflow polish, while teams see editor and licensing edge cases wo...
Agentic AI in backend systems: where autonomy wins (and where it breaks)
Agentic AI is ready to run multi-step backend workflows, but it only pays off when you bound autonomy and design for reliability. Agentic workflows fo...
Copilot model guide lands as VS Code 'Edit' mode disappears and quotas raise questions
Microsoft shared concrete guidance on choosing Copilot models while users report the Edit mode missing in VS Code and raise concerns about premium req...
Copilot model selection guidance with quota and UI gotchas
Microsoft outlines how to choose Copilot models by task while users report quota friction and a missing Edit mode after recent updates. A Microsoft gu...