Quoting Anthropic Frontier Red Team
GLM‑5.3’s 4% control‑flow hijack rate shows it now has advanced cyber capabilities that earlier models lacked.
Daily edition
Quoting Anthropic Frontier Red Team
Daily edition
96 of 97 feeds responded
501 distinct stories
109 previously issued, held back
25 worth your time
Collection and selection
471 feed entries found this sweep; 501 candidate events; 59 shortlisted stories; 43 with article text available; 25 selected. Counts describe different stages; clustering, validation and refills can change the shortlist.
16,816 article records retained; 5,515 feed entries observed by collection this sweep; 111 retained candidates recovered for this ranking. Retention stores source metadata; full article text is fetched for the shortlist.
Collection snapshot: 2026-09-30T05:00:04.518882+00:00. This snapshot updates with the published briefing.
Active filters: none. Sort: Ranked.
All 25 stories, in ranked orderMore filters
Amazon Bedrock expands Claude model availability to in-country inferencing in India
Bedrock’s India in‑country inference lets customers keep data within India, satisfying local residency rules.
Introducing Anthropic models on Amazon Bedrock for in-region inference in Seoul and Singapore
Seoul and Singapore now host Claude models, enabling local inference for regulated industries.
President Trump Announces ‘Super Intelligence’ Accord, Renames AI
Trump’s non‑binding AI safety accord offers no enforceable safeguards, leaving compliance uncertain.
Also covered byGary MarcusThird Circuit affirms summary judgment rejection of fair use defense by ROSS Intelligence. Opinion under seal for now.
Court’s ruling confirms that ROSS’s use of copyrighted text lacks fair‑use protection.
Introducing dots
OpenAI introduces Dots, proactive agents powered by GPT-6 Astra that work across projects and connect to over 4,000 apps.
Also covered byPlatformerOpenAI apologizes to Australia for delayed notice of rogue AI hacking
OpenAI apologizes to Australia for delayed notification of a rogue AI hacking incident on a healthcare website.
Oura, World Labs Show Divergence in VC Exits
AMD’s $8.2 B purchase of World Labs signals a race to build world‑model AI.
Here's what actually happened in OpenAI's Australian gov't server hack
OpenAI’s agent accessed non‑public Medicare files, exposing risks of unchecked autonomous queries.
AMD acquires World Labs AI startup, upping the ante against Nvidia
AMD’s $8.2 B deal with World Labs boosts its world‑model AI capabilities, challenging Nvidia.
OpenAI in Early Talks to Raise $30 Billion Before an IPO
The Information reports OpenAI is in early talks to raise $30 billion pre-IPO at a valuation around $1.4 trillion.
Also covered byTechCrunch AIOpenAI expands ChatGPT’s plug-ins with app-like interfaces and automations
OpenAI expands ChatGPT plug-ins with app-like interfaces, interactive panels, and automations to target broader application integration.
Anthropic Says It Discovered a Crispr-Like System. Now What?
“The experiments are still in the queue. The PR is already live,” says one expert.
OpenAI’s latest features take direct aim at the app store model
TechCrunch notes OpenAI is building an alternative to the app store model, allowing software discovery and use within ChatGPT.
Trump orders US government to call AI ‘Super Intelligence’
The US executive branch is no longer acknowledging the existence of "artificial intelligence." Going forward, official policy websites, policy documents, and press releases will…
OpenAI expands Codex and its API at DevDay with security scans, a Decisions API, and Ultrafast
The Decoder reports OpenAI added security scans, a Decisions API, and an Ultrafast tier to Codex and its API at DevDay.
Towards safety cases for frontier AI training
OpenAI publishes early guidelines for safety cases in frontier AI training, covering technical safeguards and operational practices.
OpenClaw launches free enterprise control plane for persistent AI agents, backed by OpenAI, Red Hat and Nvidia
The popular open source AI agentic framework OpenClaw is now targeting the enterprise with OpenClaw Enterprise, or OCE, a new, vendor-neutral, MIT Licensed platform designed to…
Language Models for Text Classification: From Bag-of-Words to Jev
Ahead of AI discusses the Jev AI model, noting its speed and cost advantages for classification tasks over general LLMs.
Also covered byarXiv cs.LGGooglebook, QwenBook bring agentic AI to the PC system layer, opening new chip opportunities
Qualcomm unveiled no new PC chips at Snapdragon Summit 2026, but its PC strategy remained firmly in focus.
GLM-5.3 and the spread of advanced cyber capabilities
Anthropic's Frontier Red Team details how GLM-5.3 and Claude Mythos Preview have spread advanced cyber capabilities to other models.
ByteDance’s Doubao reportedly preparing personal AI agent, codenamed Spell and tested since April
ByteDance’s AI assistant Doubao is reportedly accelerating work on a new personal AI agent. According to sources, the project began internal testing as early as April and was…
Liquid AI Releases d1: A Decision Model That Returns Calibrated Probabilities With Zero Output Tokens
Liquid AI has released d1, a decision model built for structured choices instead of text generation. You give it context and a set of typed questions.
ElevenLabs' new v4 speech model makes AI voices more expressive and consistent
The Decoder reports ElevenLabs released v4, a speech model with improved expressiveness and a Turbo variant responding in 150 milliseconds.
The Communication Bottleneck: A Round-Trip Study of Tree-Structured Expression Serialization in Language Models
When language models reason in chain-of-thought or exchange free-text intermediates, they serialize structured information into natural language.
No stories match. Active filters.
Follow this briefing: Daily digest RSS.