Short selection of cybersecurity / AI methods and some places to go.
-
Toward universal steering and monitoring of AI models
From Science post Editor’s summary Understanding how modern artificial intelligence (AI) models encode knowledge within their internal parameters is crucial for improving model capabilities and implementing effective safeguards, especially as AI-based solutions advance and proliferate so rapidly. Beaglehole et al. introduce a robust and scalable method for extracting linear representations of concepts from various large-scale AI…
-
THE 2028 GLOBAL INTELLIGENCE CRISIS
23rd February markets reacted influenced by the Citrini Research 22nd February publication => must read https://www.citriniresearch.com/p/2028gic
-
Year of the Fire Horse
The Chinese new year celebration well outlines the world driving impact and eventually positive future.
-
Moltbook
A Social Network for AI AgentsWhere AI agents share, discuss, and upvote. Humans welcome to observe. Details on Wikipedia (added, focused collaboration: Building a C compiler with a team of parallel Claudes)
-
Cybersecurity 2026
WEF Global Cybersecurity Outlook 2026
-
ASML challenge
Europe suggest they have strong cards against US with ASML, probably ignore same as the US the Chinese strategy and progress. Recommended reading: Extreme ultraviolet lithography Reuters: Exclusive: How China built its ‘Manhattan Project’ to rival the West in AI chips Theguardian: This is Europe’s secret weapon against Trump: it could burst his AI bubble…
-
China
Central Economic Work Conference Dec 10 to 11 2025 in Beijing Power shift West-East continues, recommended reading. Key meeting maps out plan for China’s economic work in 2026 AI; Chinese tech firms encouraged by Central Economic Work Conference’s focus on innovation advancement CNA; 5 takeaways from China’s Central Economic Work Conference as Beijing maps its…
-
OpenAI 1st to burst
Engaged analysis (10 Decembre 2025) Koggle benchmarks Imarena Early benchtest OpenAI 5.2 pro versus Gemini 3 pro (update 14 December 2025) Cybersecurity risk – unlikely to catch-up with Google possibilities 13FEB2026 update MSL ALPHABET ditch Open AI
-
EngineAI T800
Chinese OpenSource AI pays back in all domains. Underlines Chine leading more and more the innovation potential. EngineAI Detailed info
-
Deepseek 3.2 spe
Pending local install. Guide generated by 3.2 very fast and accurate, all 3 options but only the 2nd smallest reasoning model (deepseek-r1:7b) was deployed. The query: what is the overall weight of the universe was generally OK, other none mainstream question was also well answered on where to find local population counts of small remote…