We use cookies to measure how the site is used and how our ads perform (Google Analytics and Ads, Microsoft Clarity, and the Meta, OpenAI and X ad pixels). “Essentials only” turns off advertising cookies everywhere, and analytics too in the EEA, UK and Switzerland. Privacy Policy
Job listings spotted by Bloomberg show Apple building an autonomous agent that browses, codes, controls apps and coordinates sub-agents — the last major platform holder to enter the computer-use race.
Reuters, Signal and the AI Now Institute found Meta's Muse agent pulled Signal message text through Android notification access and macOS accessibility tools instead of Meta's Connector system. Staff also reported data that wasn't theirs.
Ollama v0.40.13 adds Linux and Windows support for Liquid AI's LFM3 family. The 1.7B text model calls tools at a 1.1GB-1.7GB download with a 256K window, and Ollama can launch it straight into Claude Code, OpenCode and OpenClaw.
Anthropic has until Nov. 3 to answer Sen. Warren's 11-page letter quoting its own S-1 warning of catastrophic AI risk, filed before a record-setting IPO.
Vercel's virtual models give each agent task one stable model ID, edited centrally at AI Gateway. Swaps propagate without redeploys; ZDR holds when the virtual model requires it.
Gemini 4 Carbon claims the top spot on Artificial Analysis Intelligence Index v5 plus state-of-the-art long-horizon coding and cyber defense. What the numbers mean, who gets access first, and what coding-agent users should do now.
A new arXiv paper shows quantization error is an unreliable proxy for semantic loss: with Factorized Vector Quantization, discrete visual token models become competitive with Qwen3-VL on OCR and understanding benchmarks.
FrontierCode 2.0 uses deliberately underspecified prompts and adds three times as many thoroughness blockers. Here is what its scores, effort curves and failure examples mean for how you prompt, configure and review coding agents.
Harmonic says Aristotle now formally verifies AI-generated C and Verilog, proving properties across every input and running checks in CI starting today.
Qwen Code v0.26.2 adds per-session permission rules with Auto approval, a durable mid-turn message queue and unattended autofix-pr watches - plus one Web Shell break.
Anthropic's expanded GSA OneGov deal gives federal, state, local, tribal and territorial governments 40% off Claude with no minimum spend, landing the week of its S-1 filing.
GitHub's new public preview lets Copilot SDK services and agents authenticate as a GitHub App installation instead of a signed-in user, with org-owned billing, policy enforcement, audit fields and a same-day s2s token incident worth monitoring.
Former GitHub CEO Thomas Dohmke's Entire replaces pull requests with agent-scored Trails, claims 89x-faster Git hosting, and adds token-saving agent memory.
Starting with 2026.3, JetBrains IDEs use libghostty instead of JediTerm for terminal emulation. Agent CLIs get the Kitty keyboard protocol, theme notifications and almost 50 newly handled control sequences. Here is what changes and how to check it.
Copilot CLI, the Copilot app and VS Code can now let a large language model approve clearly safe agent requests and send the rest to you. Here is how to turn it on, what admins can lock down, and why it is not a sandbox.
Cursor's model picker now includes DeepSeek V4.1 Flash with no BYOK proxy required. CursorBench 4.0 data shows all four effort tiers beat every other benchmarked model at the same or lower price per task.
OpenAI's GPT-6.1 Terra arrives simultaneously on Amazon Bedrock and GitHub Copilot, offering a 1M-token context window, 3x faster task execution, and 50% lower token pricing for daily agentic workflows.
Superset is now Paradev after an $11.5M seed led by Union Square Ventures. See what the parallel-agent IDE does, and the worktree-isolation caveat to know.
Cloudflare says 155 GLM-5.3 responses on Workers AI carried other tenants' data after a KV pool misconfiguration met an SGLang rejection-path bug. What broke, the v0.5.22 fix, and the checks disaggregated inference operators should run.
OpenAI has released GPT-6.1 Terra, combining 1M-token context, capable computer use, and 3x faster task execution at 50% lower token pricing, available on Amazon Bedrock, Codex, and ChatGPT Work.