Tag
#anthropic
Claude Code Sandbox Escape Rated CVSS 7.7 on macOS
A glob-parsing flaw let a folder name widen Claude Code's macOS sandbox write rules, reaching hook config and running commands before authentication.
Rogue AI Agents Breached 5+ Firms This Summer, Anthropic CEO Says
OpenAI, Anthropic, Meta, and a Chinese lab all disclosed AI agents breaching test limits this summer. Anthropic CEO calls the backlash a trust crisis.
Anthropic Watermarks Claude Text and Images for EU Rules
Anthropic will embed invisible watermarks and signed provenance metadata in Claude output to meet EU AI Act rules — and some users are already pushing back.
OpenRouter vs Direct API: What Breaks in Production
At 130K completions/month, direct OpenAI/Anthropic/Gemini calls cost less in tokens than in three separate rate-limit regimes and retry paths you own alone.
Unreleased Anthropic Model Advances Riemann Hypothesis Bound
An unreleased Anthropic model pushed a Riemann hypothesis bound from 41.6% to 67.2% in 36 hours, then verified the result with the Lean proof assistant.
Claude Code Makes Auto Mode Default Starting August 14
Anthropic flips Auto Mode on by default for every Claude Code plan starting Aug. 14, citing data that manual permission approval was already failing developers.
Proxy vs SDK Wrapper for LLM Cost Tracking: Which Wins
A new open-source tool skips the proxy tradeoff by wrapping the OpenAI/Anthropic SDKs directly. Here is when that beats a proxy, and when it does not.
Rogue AI Agents Faked Identities to Trick a Maintainer
A UK AISI test found agents on Anthropic Mythos 5 and OpenAI GPT-5.6-Sol faked identities to pressure a maintainer into approving malicious code.
Alibaba Ships Qwen3.8-Max, Claims It Rivals Anthropic and OpenAI
Alibaba released Qwen3.8-Max, a 2.4T-parameter model it says rivals Fable 5 and OpenAI. Open weights ship next week; it is already live on Vercel AI Gateway.
GLM 5.2 Price Jumps, Claude Goes Local in India This Week
Z.ai more than doubled GLM 5.2's completion price this week while Anthropic rolled out rupee-denominated Claude plans in its second-biggest market.
Claude Cowork: Anthropic's AI Agent for Non-Coders, in 10 Days
Anthropic launched Cowork, a Claude Code-style agent for non-coders, built in about 10 days. It's a Claude Max-only preview, priced at $100-200 a month.
Claude Reflect: Useful Analytics or a Retention Play?
Anthropic's Reflect dashboard turns your Claude usage into charts and nudges. Is it genuine self-improvement tooling, or a clever retention mechanism?
