Eliezer Yudkowsky Explains Why Clever Human-Designed Schemes Fail at ASI Alignment
Eliezer Yudkowsky, a leading researcher on artificial intelligence and technological singularity issues and a central figure in the LessWrong rationalist community, has released the essay Cat-Belling Problems. The text originated in 2021 and was finalized after his August 2026 debates with Jeffrey Irving at the ILIAD-2026 conference on theoretical and mathematical problems of AI safety and control.
The essay opens with the story of Michael Lorrey and his Lorrey Drive, a proposed reactionless engine discussed on the 1990s Extropians mailing list. The design featured masses rotating inside a cylinder, accelerating on one side and decelerating on the other to produce net thrust. Participants asked for explanations because the concept appeared to violate conservation of momentum and Newton's third law.
Lorrey presented an animated diagram and later claimed an Excel spreadsheet proved net force after adding secret patented compensators. He refused to show the spreadsheet. Yudkowsky notes that even if the inventor were a Nobel laureate, the claim would remain impossible under known physics because momentum is conserved in every individual interaction between nucleons, electrons, and other particles.
The deeper reason, Yudkowsky explains, stems from quantum mechanics and Noether's theorem rather than a mere surface-level summary of past observations. Any model built from standard components will sum to zero net momentum change; an arithmetic series of zeros remains zero. Claims that rely on uninspectable compensators therefore contradict mathematics, not just empirical habit.
Yudkowsky draws the direct parallel to proposals for aligning artificial superintelligence. When someone presents an intricate human-designed plan with hidden mechanisms that supposedly solves the superalignment problem, the same epistemic standard applies. The burden is not met by asserting that the plan is clever or that the details are proprietary.
Carl Feynman is quoted recalling his father Richard Feynman's classification of perpetual-motion machines and the 1970 public demonstration that ended when the power cord was unplugged. These historical cases reinforce that experimental proof, not internal spreadsheets, is required when fundamental conservation laws appear to be broken.
Related articles
Troubleshooting AI Bug Hunting: Seven Critical Mistakes That Destroy Report Validity
Two Russian researchers shared detailed lessons from six months of running AI agents on domestic bug bounty platforms. The presentation, originally delivered at OFFZONE, covers hallucinations, lack of validation, threat model misunderstandings, severity inflation, bloated reports, high token costs, and scope violations. Vladislav Bek and Daniil Shushakov describe practical fixes including tool-enabled agents, dual-agent verification, strict templates, and cascaded model architectures. They emphasize that AI must execute real commands in isolated environments rather than fabricate results. The speakers also address scaling on private events, handling model refusals, and the continued necessity of human final validation. Their checklist stresses business context, proven impact, and staying within program boundaries to produce accepted, high-value reports.
SOC Forum 2026 Reveals Full Program Featuring Chen Qiufan on AI Agents and Business Defense
Organizers of SOC Forum 2026 have published the complete agenda for the event scheduled for October 27-28 at Moscow's Lomonosov cluster. The central theme explores how generative AI is used by both attackers and defenders, including the risks when autonomous agents gain access to corporate systems. Over 170 speakers from more than 90 organizations will participate in more than 60 sessions and 82 technical reports selected from over 400 submissions. The first day focuses on business impact, regulation, economic effects of AI, liability for errors, and employee data sent to external models, with Chinese author Chen Qiufan delivering a keynote and joining a plenary discussion. The second day shifts to technical topics such as AI-powered attacks, code security, cloud environments, and autonomous agent protections including access rights, prompt attacks, and incident investigation. The AI & Cybersecurity track received 126 CFP submissions, 2.5 times more than the previous year. The forum is organized by GK Solar as part of Russian Cybersecurity Week.
Skycode: Custom VS Code Extension Optimizes AI Coding Agent for Corporate Closed-Loop LLMs
A developer has released Skycode, an open-source VS Code extension that enables AI-assisted coding inside corporate closed environments using internal models such as Qwen. The project addresses strict data residency requirements in sectors like energy, where external services like GitHub Copilot or Cursor are prohibited because code would leave the perimeter. Key engineering efforts focused on aggressive context preservation within the model's roughly 100k token limit through incremental file reading, symbol-level extraction, history compression, and visual context monitoring. Additional features include tab-based parallel task handling, local voice dictation powered by Sber's GigaAM v3, and safeguards against repetitive loops or truncated file writes. The extension approach was chosen over forking the entire VS Code codebase to simplify deployment and compliance reviews, as extensions can often be installed from the Marketplace without separate approval. Skycode is positioned as a practical alternative for organizations running their own LLMs rather than relying on cloud models.
NICT and MITRE to Host AI Breach Challenge 2026 on Email Agent Security
The National Institute of Information and Communications Technology (NICT) and MITRE are organizing the AI Breach Challenge 2026: Email Agent Security competition in December 2026. The event will take place at George Mason University in Virginia and will focus on identifying security weaknesses in AI agents designed for workplace email operations. Participants will work in teams of two to four people, combining on-site and remote access over a 24-hour period on December 16 and 17. They will attempt to induce unintended behaviors in the agents and gain access to protected information within an isolated environment. NICT plans to analyze the results to evaluate the effectiveness and limitations of current AI agent security measures for future research and development.