<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Sergey Skabelkin - FinTech, Blockchain &amp; AI Strategy</title><description>Essays on AI, fintech, digital banking, blockchain infrastructure, and regulated product strategy.</description><link>https://skabelkin.com/</link><item><title>When Models Stop Thinking Left to Right</title><link>https://skabelkin.com/en/blog/ai-agents-compliance/</link><guid isPermaLink="true">https://skabelkin.com/en/blog/ai-agents-compliance/</guid><description>Text diffusion models may weaken one of AI safety&apos;s familiar assumptions: that reasoning can be monitored as a sequential chain of thought.</description><pubDate>Mon, 22 Jun 2026 00:00:00 GMT</pubDate><category>AI Safety</category><category>Reasoning</category><category>Diffusion Models</category></item><item><title>Что будет с AI safety, если модели перестанут мыслить последовательно</title><link>https://skabelkin.com/ru/blog/ai-agents-compliance/</link><guid isPermaLink="true">https://skabelkin.com/ru/blog/ai-agents-compliance/</guid><description>Почему текстовые диффузионные модели могут сломать привычную идею мониторинга цепочки рассуждений и потребовать новых подходов к AI safety.</description><pubDate>Mon, 22 Jun 2026 00:00:00 GMT</pubDate><category>AI Safety</category><category>Reasoning</category><category>Diffusion Models</category></item><item><title>From Balls to Consciousness: Testing Natural Abstractions</title><link>https://skabelkin.com/en/blog/digital-banks-zero-to-market/</link><guid isPermaLink="true">https://skabelkin.com/en/blog/digital-banks-zero-to-market/</guid><description>Alignment theories need more than elegant mathematics; they need test cases that range from simple objects to concepts like monogamy and consciousness.</description><pubDate>Sun, 21 Jun 2026 00:00:00 GMT</pubDate><category>AI Alignment</category><category>Concepts</category><category>Natural Abstractions</category></item><item><title>От мяча к сознанию: как проверять natural abstractions</title><link>https://skabelkin.com/ru/blog/digital-banks-zero-to-market/</link><guid isPermaLink="true">https://skabelkin.com/ru/blog/digital-banks-zero-to-market/</guid><description>Почему alignment-теориям нужны не только красивые математические рамки, но и тестовые наборы концептов от простых объектов до сознания.</description><pubDate>Sun, 21 Jun 2026 00:00:00 GMT</pubDate><category>AI Alignment</category><category>Concepts</category><category>Natural Abstractions</category></item><item><title>Safety, Agency, and the Difference Between Constraint and Choice</title><link>https://skabelkin.com/en/blog/rwa-tokenization-regulated-market/</link><guid isPermaLink="true">https://skabelkin.com/en/blog/rwa-tokenization-regulated-market/</guid><description>For agentic AI systems, there is a crucial difference between being unable to cause harm and consciously choosing not to.</description><pubDate>Sat, 20 Jun 2026 00:00:00 GMT</pubDate><category>AI Safety</category><category>Agency</category><category>Governance</category></item><item><title>Безопасность слабости и безопасность силы</title><link>https://skabelkin.com/ru/blog/rwa-tokenization-regulated-market/</link><guid isPermaLink="true">https://skabelkin.com/ru/blog/rwa-tokenization-regulated-market/</guid><description>Почему для агентных AI-систем важно различать неспособность причинить вред и осознанный отказ от вреда.</description><pubDate>Sat, 20 Jun 2026 00:00:00 GMT</pubDate><category>AI Safety</category><category>Agency</category><category>Governance</category></item></channel></rss>