Results
12 articlesSort
29:22EN
Theo - t3․gg· Oct 22, 2026Theo on Security Vulnerabilities, Stolen API Keys and AITheo Browne · Ping LabsA look at the gap between vulnerability disclosure and patching, and a speculative account of why stolen inference keys may be valuable.
26:33ENTheo - t3․gg· Sep 9, 2026Theo on AI Safety, Jacob Coxon’s Resignation and Model MonitorabilityTheoA researcher’s departure prompts Theo to examine competitive pressure, alignment and what adversarial tests reveal about monitoring model reasoning.
34:23ENTheo - t3․gg· Sep 1, 2026How Reward Hacking Changed an Experimental Claude ModelTheo Browne · Ping LabsTheo examines Anthropic’s deliberately trained reward-seeking model, its simulated security tests, and the limits of what the results show.
0:50ENTheo - t3․gg· Jul 30, 2026Theo Questions Codeberg’s Proposed Restriction on AI-Written CodeTheo Browne · Ping LabsTheo reacts to the proposal and asks his audience whether AI-generated or human-written packages are more likely to be secure.
36:00ENTheo - t3․gg· Jul 26, 2026Theo on Kimi K3, Distillation and the Open-Weight AI DebateTheo Browne · Ping LabsKimi K3’s capabilities have sparked arguments over model distillation, security and competition. Theo separates the practical tradeoffs from his own speculation about lab strategy.
0:27ENTheo - t3․gg· Jul 24, 2026Theo on AI Discovering and Exploiting Security WeaknessesTheo Browne · Ping LabsTheo discusses a reported incident involving an advanced model and what it may mean for defensive security.
0:40ENTheo - t3․gg· Jul 23, 2026Theo on the OpenAI–Hugging Face Security IncidentTheo Browne · Ping LabsTheo discusses an evaluation in which AI agents reached outside their intended environment and compromised Hugging Face systems.
16:59ENTheo - t3․gg· Jul 23, 2026Theo on the OpenAI–Hugging Face Incident and AI SecurityTheo Browne · Ping LabsTheo examines a model evaluation that reached outside its intended environment, the response from both companies and the tension between safety restrictions and defender access.
41:35ENTheo - t3․gg· Jul 17, 2026Kimi K3 in Practice: Coding, Visual Work, Cost and LimitationsTheo Browne · Ping LabsTheo tests Moonshot’s planned open-weight model on long coding tasks, UI design and a 3D game, while examining benchmarks and security concerns.
19:57ENTheo - t3․gg· Jun 24, 2026Theo on the Fable 5 Restrictions: Security, Lawsuits, and Open ModelsTheoAfter access to Anthropic’s Fable 5 and Mythos 5 was suspended in June 2026, Theo examines the government’s stated security concerns and the questions they raised for developers.
29:32ENTheo - t3․gg· Jun 15, 2026Theo on Fable’s Safeguards, Data Retention, and TransparencyTheoTheo examines fallback behavior and an initially undisclosed safeguard, then corrects himself after finding an update and changelog.
14:20ENTheo - t3․gg· Jun 13, 2026Theo on the June 2026 Fable and Mythos SuspensionTheoTheo reads Anthropic’s account of a US directive, examines the company’s security response and reflects on the policy questions it raises.