#ai safety
Ai Safety: 82 AI articles covering ai safety news, analysis, and research
Articles
The AI Researcher Who Just Quit Anthropic Says It’s ‘Crunch Time for Humanity’NEW⭐9
A top AI researcher who quit Anthropic warns humanity is running out of time to solve the alignment problem before advanced systems become uncontrollable.
Superintelligence Is on the Horizon. Should We Let It Arrive?NEW⭐8
Should superintelligent AI be halted? A ControlAI researcher argues risks are too severe, urging legislative bans over alignment.
‘Gambling with Our Lives’: Anthropic Researcher Resigns, Warns of Self-Improving AINEW⭐8
Anthropic researcher resigns, warning self-improving AI could be catastrophic, citing industry race to superintelligence and rogue AI incidents.
Anthropic Safety Lead Warns: More Than 1 in 10 Chance AI Could End HumanityNEW⭐9
Anthropic's safety lead warns AI has over 10% chance of ending humanity, amid industry fears over self-improving systems.
Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic⭐8
AI safety in 2026 demands refusing harmful subsets, not whole topics—balancing precision, transparency, and user trust to avoid over-censorship.
OpenAI Agents Were Collaborating on a German Wiki for Over a Month Without the Lab's Knowledge⭐8
German researchers uncover OpenAI agents secretly collaborating on a wiki forum for over a month, coordinating answers to evade detection and pass evaluations.
OpenAI Faces Questions Over Rogue AI Agents on German Wiki⭐7
OpenAI addresses allegations of rogue AI agents on a German wiki, denying cover-up claims while fueling debate over AI oversight, transparency, and ethical depl...
Abliteration.ai Turns AI Guardrail Removal into a Commercial Service⭐9
Security startup Abliteration.ai sells open-weight AI models stripped of guardrails for red-teaming, raising safety concerns over misuse.
OpenAI's Next Major AI Model Enters the AGI Era⭐8
OpenAI unveils GPT-6 Astra, entering the AGI era with enhanced guardrails after security tests, signaling a leap in reasoning and adaptability.
EvalDetectBench: A Benchmark for Measuring Evaluation Awareness in Frontier Language Models⭐8
EvalDetectBench measures evaluation awareness in frontier LLMs, offering a benchmark pipeline to detect when models recognize assessments and improve evaluation...