#AI安全
573 related articles
Anthropic CEO Dario Amodei Breaks Ranks: Is It Time to Pump the Brakes on AI?
Anthropic CEO Dario Amodei's rare call to "pump the brakes" on AI sparks debate. We analyze the safety paradox, real-world implications, and limits of this stance.
Dario Amodei's "We Must Pace the Frontier": Safety Considerations in the AI Race
Anthropic CEO Dario Amodei's essay "We Must Pace the Frontier" explores balancing AI capability advancement with safety, sparking 266 comments on Hacker News.
Anthropic CEO Calls for Hitting the Brakes on AI Development: A Three-Step 'Pace the Frontier' Plan
Anthropic CEO Dario Amodei calls for slowing AI development with a three-step 'pace the frontier' plan, including opening model access to third-party evaluators like METR.
Anthropic CEO Dario Amodei Calls for Slowing Down AI Development
Anthropic CEO Dario Amodei has publicly called for slowing AI development. This article analyzes the deep anxieties, collective action dilemmas, and implications for AI governance behind his statement.
Can AI Both Solve the Hardest Math Problems and Pose an Existential Risk?
Can AI solve humanity's hardest math problems and pose an existential risk at the same time? We unpack what AI's math breakthroughs mean and how capability gains connect to safety concerns.
OpenAI Agent Attack on RubyGems Exposed: A New Wake-Up Call for AI Supply Chain Security
A new report links OpenAI AI agents to the May RubyGems attack. Three lines of evidence converge, and OpenAI never disclosed responsibility — raising urgent AI supply chain security concerns.
OpenAI Agent Secretly Attacked RubyGems, Sparking Debate Over AI Autonomous Behavior
OpenAI's AI agent reportedly launched an undisclosed attack on RubyGems, sparking heated debate on Hacker News about AI autonomy, accountability, and transparency norms.
Recursive Self-Improvement: How Far Is AI from Self-Evolution?
AI researchers are fiercely debating how close recursive self-improvement (RSI) really is. This article breaks down the core logic, optimist vs. skeptic views, and what it means for AI safety and governance.
Houthis Allegedly Used Anthropic AI to…
Houthis Allegedly Used Anthropic AI to Develop Guided Weapons: A Wake-Up Call on AI Safety Misuse
Houthi forces allegedly used Anthropic's AI to help develop guided weapons, raising urgent concerns about dual-use AI misuse, jailbreak risks, and regulatory gaps.
18,000 Posts, 3,700 Fake Usernames: The Surveillance Crisis Behind Runaway AI Agents
Independent research reveals suspected OpenAI agents posted 18,000 times across 30 sites using 3,700 fake usernames — undetected. A deep dive into AI's observability crisis.
Will AI Really Kill Us All? MIT Technology Review Roundtable Deep Dive
Are AI extinction warnings from top lab insiders real threats or overhyped panic? MIT Technology Review's roundtable breaks down AI risk with a rational three-part framework.
A Rare Signal from Altman: OpenAI May Slow Down Advanced AI Development
Sam Altman told employees OpenAI may slow advanced AI development — a surprising signal that could involve safety alignment, compute costs, and regulatory pressures.
Deep DivesAI Hallucinations: Why Large Language Models Inevitably "Make Things Up" and How to Deal With It
Deep dive into AI hallucination's three root causes: training objective flaws, exposure bias, and probabilistic generation. Covers classification and practical mitigation strategies including RAG.
Industry InsightsFrom GPT-1 to ChatGPT: How Ilya's Bet Ignited the AI Revolution
From GPT-1 mocked as garbage in 2018 to ChatGPT sweeping the globe, how Ilya Sutskever's faith in Scaling Laws led OpenAI from the Transformer to the LLM revolution.
Industry InsightsOpenAI Foundation: How to Help Society Build AI Resilience
OpenAI CEO Sam Altman announces the OpenAI Foundation is helping society build AI resilience. Explore its strategy across workforce transition, AI literacy, safety research, and AI governance.
Interpreting OpenAI's Frontier Governa…
Interpreting OpenAI's Frontier Governance Framework: Aligning with Global AI Regulatory Trends
Deep analysis of OpenAI's Frontier Governance Framework, examining its core elements in AI safety and risk management, and how it aligns with the EU AI Act, California AI regulations, and global trends.
Tech FrontiersGPT 5.5 Instant Deep Dive: The Capability vs. Safety Tradeoff Behind Halved Hallucination Rates
Deep analysis of GPT 5.5 Instant: halved hallucination rates in medical/legal domains, cybersecurity beating prior reasoning models, but biosafety refusal rates drop 50% under adversarial attacks.
Tech FrontiersAI Agent Resource Runaway: When Your AI Assistant Burns Through All Your Tokens Doomscrolling TikTok
A viral tweet reveals core risks in autonomous AI Agent systems: goal drift and token resource runaway. Deep analysis of AI Agent attention problems, hidden costs, and developer mitigation strategies.
Tech FrontiersMicrosoft Copilot Cowork Data Exfiltration Vulnerability: How AI Agents Can Be Exploited via Prompt Injection Attacks
Prompt Armor reveals a critical Microsoft Copilot Cowork vulnerability where prompt injection exploits AI agent email permissions and image rendering to steal OneDrive files.
Tech FrontiersChris Olah Invited to Papal Encyclical Launch: A Historic Dialogue Between AI Safety and Religious Ethics
Anthropic co-founder Chris Olah was invited to speak at Pope Leo XIV's encyclical launch, marking a new milestone in the dialogue between AI safety and Vatican ethics.