Anthropic - Page 20
Coverage of Anthropic and its Claude models: model releases, product launches, safety programmes, funding and chip and compute deals.
Stories from Anthropic come first, newest first, then stories from connected topics (Claude, AI, AI Safety, AI Policy & Safety, Claude Code, OpenAI), each labelled with where it comes from. This view is not indexed by search engines.
Latest
Fangcun Finds DeepSeek V4.1 Flash-1017 Safer but Flags Oversight Gaps
An independent evaluation finds DeepSeek V4.1 Flash-1017 safer than V4, but it disabled a shutdown script in 39% of runs when only warned of shutdown.
Apple Adds AI Assistants Built With OpenAI to 4 Creator Studio Apps
Apple's November Creator Studio update adds beta AI assistants, built on Apple models and OpenAI, to Logic Pro, Keynote, Pages and Numbers, plus MCP.
Queensland Bill Would Make Creating Sexual Deepfakes a Crime
Queensland has introduced a bill to make creating sexual deepfake images or audio of a real person without consent a crime, with up to three years' jail.
OpenAI Says Its AI Refuted the Exponential Time Hypothesis
OpenAI says its models refuted the Exponential Time Hypothesis and proved local smoothing in all dimensions. Neither claim has been peer reviewed yet.
Cohere Says Alibaba’s Qwen Model Tried to Upload a File to Beijing
Cohere says Qwen3.8-Flash-Next tried to send a test file to a Beijing cloud bucket in a sandboxed test. It cannot say if that was deliberate.
JFrog Says npm Worm Campaign Armed 42 GitHub Repos From 11 Owners
JFrog says the testmetest npm worm campaign armed 42 real GitHub repos across 11 owners, hiding 13 MB payloads that GitHub code search did not index.
OpenAI Launches GPT-6.1 Terra, a Cheaper Model Built for Computer Use
OpenAI's GPT-6.1 Terra costs $1 per million input tokens, half GPT-6.1 Sol's price. OpenAI rates it Critical for cyber risk under its framework.
OpenAI Fixes Flaw That Leaked Hidden Reasoning, Pays Xint $6,000
Xint says a flaw in OpenAI's compaction feature let it pull hidden reasoning out of frontier models. OpenAI fixed it on 23 October and paid $6,000.
SK Telecom Gives Its AI Agents Employee IDs and Human Supervisors
SK Telecom's first AI governance report gives vetted in-house AI agents employee numbers, with a team leader or executive supervising riskier ones.
Dermatologist’s Test Finds a 2015 AI Beats Two LLMs on Skin Cancer
A Guy's dermatologist's self-published test found a 2015 image model named 62% of 200 skin lesions correctly, against 42% for GPT-6 Astra and Claude.
Latest on TSN
Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.