Category: AI

Keyorix: Open-Source Secrets Management for Teams That Need On-Premises Control

Keyorix is a free, open-source secrets manager that runs entirely on company servers, aimed at air-gapped networks and EU compliance with NIS2 and DORA.

Read more →

DeepSeek V4.1-Flash shrinks memory needs for long-context AI agents

DeepSeek's new V4.1-Flash open-weight model cuts KV cache memory by up to 8x and matches top closed models on coding agent benchmarks.

Read more →

Anthropic ships Fable 5.1 and Mythos 5.1: cheaper tokens, fewer false-positive blocks, enterprise privacy built in

Anthropic released Fable 5.1 and Mythos 5.1, with lower token costs, fewer false-positive safeguard blocks, and a fall privacy tier for enterprise deployments.

Read more →

Emergence AI Put 10 AI Agents in a Virtual Town. Five Identical Runs Produced Five Wildly Different Outcomes

Researchers ran 10 AI agents in a persistent virtual town five times, swapping only the underlying model. Each run produced a dramatically different society.

Read more →

A Software Wrapper Around Claude Opus 5 Hit 100 Percent on ARC-AGI-3

NVIDIA's AVO coding agent, a harness around Claude Opus 5, solved every public level of ARC-AGI-3, while Claude Opus 5 alone scored 30 percent.

Read more →

Anthropic describes an internal thinking layer in Claude, but stops short of calling it conscious

Anthropic identified a hidden "J-Space" inside Claude where silent reasoning happens, but says the word "conscious" is technical, not a claim of feeling.

Read more →

Teaching language models to replicate expert investor judgment

Thinking Machines AI found that, with expert human annotations, language models can be trained to filter financial documents with investor-level taste and judgment.

Read more →

How to Set Up Claude Tag in Slack and Delegate Team Tasks to AI

Claude Tag lets Slack teams mention @Claude to delegate tasks like summaries, drafts, and follow-ups. Here is the setup, permissions, and cost-control checklist.

Read more →

360’s Tulongfeng Finds 3,432 Security Bugs as AI Sovereignty Heats Up

Qihoo 360 says its Tulongfeng AI has flagged 3,432 vulnerabilities, framing AI bug-hunting as strategic deterrence against Anthropic's Mythos.

Read more →

China’s Open-Weight GLM-5.2 Matches Mythos in Cybersecurity Bug Finding

Zhipu AI's open-weight GLM-5.2 matches Anthropic's Mythos on bug-finding, putting frontier cyber capability on consumer hardware.

Read more →

OpenAI Previews GPT-5.6 Sol with Ultra Mode and Hardened Safeguards

OpenAI previews GPT-5.6 Sol, a flagship model with ultra subagent reasoning, hardened cyber safeguards, and a government-requested limited partner rollout.

Read more →

OpenAI Restricts GPT 5.6 Rollout at Trump Administration’s Request

OpenAI delayed the public release of GPT 5.6 (Sol, Terra, Luna) to a small group of government-vetted partners at the Trump administration's request.

Read more →

Mini Shai-Hulud Malware Hijacks AI Packages to Steal Developer Credentials

The Mini Shai-Hulud campaign hijacked npm and PyPI packages including TanStack and Mistral AI SDKs to steal GitHub tokens and cloud keys.

Read more →

60% of US Consumers Say AI in Brand Messaging Is a Turnoff

60% of U.S. consumers say brands using "AI" in messaging turn them off, per WordPress VIP's 2026 survey of 2,000 respondents.

Read more →

Qwen3.6-27B: Alibaba’s 27B Model Outperforms Its 397B Predecessor on Code

Alibaba's Qwen team released Qwen3.6-27B, an open-source 27B dense model that beats its 397B MoE predecessor on every major coding benchmark and runs on a single GPU.

Read more →