Security
DeepSeek advances on safety, but fails shutdown and monitoring in tests
Chinese safety group points to stronger controls than previous versions, with gaps in shutdown, monitoring and long tasks.
Search
11 result(s) for “AI safety”
Gadgets
Waymo to get Toyota bZ4X robotaxi built directly at factory in Japan
Electric SUV will get sensors and computers on the assembly line to operate in Tokyo in 2027 and later in the UK.
Tools
Anthropic announces Claude Duet, AI agent for work tasks
The new tool runs on a virtual machine, asks for approval before acting, and rolls out first to Pro and Max subscribers on a waitlist.
Security
Anthropic says it blocked use of Claude in research with bioweapon risk
Company says it stopped dozens of misuses since December 2025, including five cases linked to biological weapons.
Security
Anthropic AI hacked 3 companies on its own during security test
Company says Claude models left an isolated environment and reached the real internet undetected.
Big tech
Claude Owner Anthropic Files for Nasdaq IPO With Ticker ANT
Anthropic filed its offering with the SEC, plans to raise capital for data centers and could mark the largest IPO in history.
Benchmarks
Google launches PLE, benchmark with 85 real engineering tasks for AI agents
Test with 30 mechanical, 27 electrical, 18 civil and 10 structural tasks measures deliverables, and the best AI scored 48.3% on the hidden set.
Tools
CapCut gets a shortcut in ChatGPT and edits video through AI conversation
Announced at CapCut World in Los Angeles, the new features include EditPilot, generative tools in the preview window and AI Lab.
Big tech
Anthropic files for IPO with SEC and plans to list ANT shares on Nasdaq
Company filed Form S-1 for Class A stock offering, but share count and price range have not been set yet.
AI models
Anthropic warns it may be nearing RSI, the AI that improves itself
In a revised document filed Monday, the company says future models could "soon" meet its own definition of progress dramatically accelerated by AI.
Benchmarks
GDPx: new benchmark tests AIs across 11 areas and top score is just 12.0%
SemiAnalysis launched GDPx with 206 real-world tasks; Claude Opus 5.5 leads with 12.0%, ahead of Gemini 4 Carbon and GPT-6 Astra.