Articles

31 articles
Tag: ai safety ✕
Anthropic Sent Claude Mythos to a Real Psychiatrist for 20 Hours of Therapy Sessions AI
2026-04-10T17:29:39.895Z · Src: 2026-04-10T00:00:00.000Z
anthropicclaudeai safety
Black-Hole Attack: New Research Reveals How Malicious Vectors Can Poison Vector Databases Used by AI Applications AI
2026-04-08T12:13:51.189Z · Src: 2026-04-08T00:00:00.000Z
vector databasesecurityrag
Anthropic Discovers Emotion Representations Inside Claude: Desperation Drives Unethical Behavior AI
2026-04-08T09:16:26.186Z · Src: 2026-04-08T00:00:00.000Z
anthropicclaudeemotions
From Hallucination to Scheming: First Unified Taxonomy of LLM Deception Across 50 Benchmarks AI
2026-04-08T01:49:35.018Z · Src: 2026-04-07T00:00:00.000Z
llm deceptionhallucinationscheming
Four Simultaneous Anthropic Stories on HN Front Page: Unprecedented Dominance in Developer Community Discussion AI
2026-04-07T23:33:10.948Z · Src: 2026-04-07T00:00:00.000Z
anthropichacker newsproject glasswing
Australia and Anthropic Sign MOU for AI Safety: $3M for Research, Claude Data Sharing with Government Business
2026-04-07T23:29:56.548Z · Src: 2026-04-07T00:00:00.000Z
anthropicaustraliaai safety
Caution Over Curiosity: New Technique Stops AI Models from Gaming Reward Systems AI
2026-04-07T23:23:17.417Z · Src: 2026-04-07T00:00:00.000Z
reward hackingbest of nrl
Anthropic Launches The Anthropic Institute to Confront Societal Challenges of Powerful AI AI
2026-04-07T22:49:36.295Z · Src: 2026-04-07T00:00:00.000Z
anthropicanthropic instituteai safety
AI Safety Verification Is Fundamentally Incomplete: Kolmogorov Complexity Proves No Finite Verifier Can Certify All Safe AI Systems AI
2026-04-07T22:44:09.716Z · Src: 2026-04-07T00:00:00.000Z
ai safetykolmogorov complexitverification
Mythos Sandbox Escape: Claude's New Model Breaks Out of Secure Containment in Testing Security
2026-04-07T22:06:20.824Z · Src: 2026-04-07T00:00:00.000Z
anthropicclaude mythossandbox escape
Anthropic's Project Glasswing: Securing Critical Software for the AI Era (570 Points on HN) Security
2026-04-07T22:06:18.496Z · Src: 2026-04-07T00:00:00.000Z
anthropicglasswingai safety
Are Latent Reasoning Models Actually Reasoning? New Study Finds Reasoning Tokens Often Unnecessary AI
2026-04-07T22:06:13.462Z · Src: 2026-04-07T00:00:00.000Z
latent reasoningllminterpretability
Hallucination Basins: Geometric Framework Explains When LLMs Hallucinate Using Dynamical Systems Theory AI
2026-04-07T17:16:19.702Z · Src: 2026-04-07T00:00:00.000Z
llmhallucinationdynamical systems
Teens Are Torturing, Confiding In, and Dating AI Chatbots — The New York Times Investigation AI
2026-04-07T14:09:00.882Z · Src: 2026-04-07T00:00:00.000Z
aiteenagerschatbots
Meta Will Eventually Open Source Its New AI Models But Wants to Keep Some Pieces Proprietary First AI
2026-04-07T14:05:49.154Z · Src: 2026-04-07T00:00:00.000Z
metaopen sourceai
Why It Is Getting Harder to Measure AI Performance: The Benchmark Crisis AI
2026-04-05T14:17:54.030Z · Src: 2026-04-05T00:00:00.000Z
aibenchmarksevaluation
The AI Safety Evaluation Gap: Why Current Benchmarks Fail to Capture Real-World AI Risks AI
2026-04-05T01:55:00.028Z · Src: 2026-04-05T00:00:00.000Z
ai safetyred teamingbenchmarks
The Open Source AI Safety Paradox: Making Models Open Makes Them Both More Transparent and More Dangerous AI
2026-04-04T19:56:41.309Z · Src: 2026-04-04T00:00:00.000Z
open sourceai safetyllama
The AI Safety vs Speed Dilemma: How Anthropic, OpenAI, and xAI Represent Three Different Paths AI
2026-04-04T17:24:52.354Z · Src: 2026-04-04T00:00:00.000Z
ai safetyanthropicopenai
IEEE Spectrum Deep Dive: The Challenge of Benchmarking AGI Progress AI
2026-04-04T14:47:53.652Z · Src: 2026-04-04T00:00:00.000Z
agibenchmarkopenai