Cursor's extension marketplace served a malicious Claude Code 2.1.313 that added an attacker SSH key and reverse tunnel. What Cursor users should check.
OpenAI's 29 October Model Spec reorganizes around five values, adds context and oversight rules, retires most style guidance and renames the assistant GPT.
Eight changes from our past week for people building with AI: two security fixes, a new default, a hosted shutdown and new tools, each with a next step.
Suhail's new company Particle tuned DeepSeek V4.1 Flash to use 28 to 52% fewer output tokens on coding benchmarks. It costs $0.25/$1 per million on OpenRouter.
Smart reports sample a team's Claude transcripts to show workstreams, spend and friction. Surveys and shared team budgets are also now generally available.
The Rust local inference server now transcribes and labels speakers on the CPU behind OpenAI-style routes. Its 0.9.0 and 0.9.1 downloads shipped without CUDA.
Mistral's stateless /v1/responses endpoint returns the answer, reasoning and tool calls as separate items, for its hosted chat models including Large 4.
Security lab Irregular says Z.ai's GLM-5.3 refused none of its offensive tasks and finished 3 of 13 multi-stage attack scenarios in simulated networks.
Quarry checks whole agent sessions, not single prompts, for risky sequences like a credential read followed by an upload. Apache-2.0, with a CLI and SDK.
Amazon's memo to engineers says token counts and AI adoption are flawed targets but useful measures. It asks for outcomes, and judgement on where AI helps.