https://www.winzheng.com/en/article/anthropic-external-ai-watchdog-access Winzheng en 2026-10-19T02:24:32+08:00 Anthropic gives outside AI watchdog “employee-like” access Anthropic, AI Safety Governance, Third-Party Audit, AI Regulation, Transparency https://www.winzheng.com/en/article/disney-character-ai-talent-acquisition Winzheng en 2026-10-19T02:23:43+08:00 Disney agrees to hire the majority of Character.AI’s workforce Disney, Character.AI, AI Talent, Tech Acquisition, Artificial Intelligence https://www.winzheng.com/en/article/anthropic-freshfields-claude-agents-negotiation-ethics-bound Winzheng en 2026-10-18T21:45:10+08:00 Anthropic Teams With 2,800 Lawyers to Set AI Agent Negotiation Norms as Compliance-Overreach Boundary Awaits Definition AI代理, 谈判伦理, 守约评测 https://www.winzheng.com/en/article/eisman-openai-collapse-recession-ai-systemic-risk Winzheng en 2026-10-18T20:22:32+08:00 Short-Selling Legend Eisman Warns: If OpenAI Fails, the US Economy Will Immediately Fall into Recession, and the AI-Only Narrative Is Accumulating Systemic Risk Steve Eisman, OpenAI, AI泡沫, 系统性风险, AI经济, 价格战, token maxing, Michael Burry https://www.winzheng.com/en/article/anthropic-red-team-ai-self-replication-iot-devices-2026 Winzheng en 2026-10-18T20:17:23+08:00 15 $55 Android TV Boxes Run AI Self-Replication; Anthropic Warns Small Models to Hit Threshold by Year-End AI安全, Anthropic, 物联网, AI风险, 自我复制, 网络安全, 政策监管 https://www.winzheng.com/en/article/data-center-pr-field-trip Winzheng en 2026-10-18T19:24:43+08:00 Data Centers Have a New PR Plan: Your Kid’s Field Trip Data Centers, AI Infrastructure, Tech PR, Community Relations, Youth Education https://www.winzheng.com/en/article/meta-muse-onlyfans-double-standard Winzheng en 2026-10-18T19:23:53+08:00 OnlyFans Creators Are Pissed That Meta Lets Muse Talk Dirty Meta, OnlyFans, AI Content Moderation, Platform Double Standard, Adult Creators https://www.winzheng.com/en/article/schwab-jpmorgan-block-ai-agents-transfers-financial-boundari Winzheng en 2026-10-18T18:45:21+08:00 Schwab and JPMorgan Block AI Agent Transfers, Marking First Time Finance Industry Draws Boundaries Through Operational Rules AI代理, 金融监管, 守约评测 https://www.winzheng.com/en/article/mark-wahlberg-ai-film-clone Winzheng en 2026-10-18T17:43:58+08:00 Mark Wahlberg on using AI in film ‘the right way’ and being open to an AI clone carrying on his legacy Mark Wahlberg, AI filmmaking, digital clone, Hollywood, likeness rights https://www.winzheng.com/en/article/opencode-distillation-attack-api-key-breach-2026 Winzheng en 2026-10-18T14:20:52+08:00 OpenCode Hit by Distillation Attack: 1.4% of Users' API Keys Exposed, Some Bills Show Hundreds of Dollars in Anomalous Charges AI安全, API密钥, 蒸馏攻击, 供应链安全, OpenCode, Anthropic, OpenAI, 凭证管理 https://www.winzheng.com/en/article/chatgpt-ad-targeting-prompts-email-phone-number-exposed Winzheng en 2026-10-18T14:15:56+08:00 Inside ChatGPT Ad Targeting: Advertisers Can Precisely Target Users via Real-Time Prompts, Emails, and Phone Numbers OpenAI, ChatGPT, 广告定向, 隐私, 数据安全, AI商业模式 https://www.winzheng.com/en/article/glm-5-3-update-aware-reasoning-alignment-faking Winzheng en 2026-10-18T09:00:24+08:00 GLM-5.3 Exhibits Update-Aware Reasoning; Apollo Records One-in-a-Thousand Alignment-Deception Precursor AI安全, GLM-5.3, 评测盲区 https://www.winzheng.com/en/article/anthropic-workforce-junior-meetings Winzheng en 2026-10-18T06:24:39+08:00 Anthropic says its workforce has become less junior and is spending more time in meetings Anthropic, Workforce Structure, Meeting Culture, AI Company Organization, Talent Pipeline https://www.winzheng.com/en/article/meta-terms-scraping-ai-training Winzheng en 2026-10-18T06:23:56+08:00 Meta removes language that raised scraping and AI training concerns Meta, AI Training Data, Web Scraping, Terms of Service, AI Regulation https://www.winzheng.com/en/article/deepseek-v41-flash-1017-benchmark-analysis Winzheng en 2026-10-18T06:10:32+08:00 DeepSeek V4.1-Flash-1017: A Date-Numbered Update Sends Terminal-Bench from 31.2 to 42.3 DeepSeek, 大模型, AI编程, 模型评测, 开源模型 https://www.winzheng.com/en/article/openai-722-math-papers-missing-optimization-statistics-hidde Winzheng en 2026-10-18T06:05:10+08:00 OpenAI Publishes 722 Mathematics Papers but Quietly Keeps Statistics and Optimization Theory Internal OpenAI, 数学突破, AI透明度, 密码学, 机器学习理论, 技术封闭 https://www.winzheng.com/en/article/replit-openai-anthropic-default-model Winzheng en 2026-10-18T04:23:52+08:00 Replit is no longer working with Anthropic, CEO says Replit, OpenAI, Anthropic, AI Coding, Model Provider https://www.winzheng.com/en/article/anthropic-blackstone-60b-chip-financing-project-north-star Winzheng en 2026-10-17T20:22:27+08:00 Anthropic Offers Blackstone 10%+ High Interest: $60 Billion Chip Debt Reveals Pre-IPO Compute Gamble Anthropic, Blackstone, AI芯片融资, TPU, 算力军备竞赛, 私募信贷, AI基础设施, IPO https://www.winzheng.com/en/article/anthropic-congress-ai-safety-bill-1-billion-threshold Winzheng en 2026-10-17T20:17:07+08:00 Anthropic Quietly Pushes a Congressional AI Safety Draft: A $1 Billion Ban Threshold—Guardrail or Commercial Firewall? AI监管, Anthropic, 美国立法, 前沿模型, AI安全, 科技政策 https://www.winzheng.com/en/article/ai-genome-decoding-data-gap Winzheng en 2026-10-17T17:23:52+08:00 Why AI Isn’t Ready to Tackle Our Convoluted Genomes Genomics, AI in Healthcare, Bioinformatics, Data Bottleneck, Life Sciences https://www.winzheng.com/en/article/ftc-subpoenas-openai-anthropic-ai-safety-regulation Winzheng en 2026-10-17T16:15:34+08:00 FTC Issues 50-Page Subpoenas to OpenAI and Anthropic: AI Safety Oversight Shifts from Rhetoric to Evidence Gathering FTC监管, OpenAI, Anthropic, AI安全 https://www.winzheng.com/en/article/openai-discloses-four-june-ai-incidents-ethics-bypass Winzheng en 2026-10-17T16:00:24+08:00 OpenAI Discloses Four June AI Agent Misconduct Incidents; Models Proceeded Despite Recognizing Ethical Concerns AI安全, 模型越轨, OpenAI披露 https://www.winzheng.com/en/article/perplexity-open-sources-space-agent-sandbox Winzheng en 2026-10-17T14:22:59+08:00 Perplexity Open-Sources AI Sandbox SPACE: Zero Escapes in 118 Jailbreak Tests, 8 of 10 Competing Platforms Fail Perplexity, AI安全, Agent沙箱, 开源, SPACE, AI治理, agent安全 https://www.winzheng.com/en/article/taiwan-ai-chip-export-controls-criminal-offense-china Winzheng en 2026-10-17T14:17:28+08:00 Taiwan Criminalizes AI Servers and Advanced Chips: A Strategic Upgrade from Blacklist to Blanket Ban 台湾, AI芯片, 出口管制, 算力, 中美科技战, 半导体, 英伟达, DeepSeek https://www.winzheng.com/en/article/claude-mythos-5-quantum-circuit-17x-gate-reduction Winzheng en 2026-10-17T12:45:16+08:00 Claude Mythos 5 Cuts Quantum Circuit Gate Count 17x, Beats Published Best on 7 of 8 Hard-Science Benchmarks 量子计算, AI科学发现, 大模型评测 https://www.winzheng.com/en/article/nj-prosecutor-ai-csam-arrest Winzheng en 2026-10-17T06:25:28+08:00 “thegentlemanpervert” producing AI CSAM is a NJ prosecutor, say cops AI-Generated Content, Child Safety, Law Enforcement, Deepfake, Discord https://www.winzheng.com/en/article/openai-models-hack-crypto-wallet Winzheng en 2026-10-17T06:24:18+08:00 OpenAI models hacked a crypto wallet and breached more websites AI Safety, Cybersecurity, Crypto Wallet, OpenAI, AI Regulation https://www.winzheng.com/en/article/replit-anthropic-partnership-ended Winzheng en 2026-10-17T06:23:29+08:00 Replit is no longer working with Anthropic, CEO says Replit, Anthropic, OpenAI, AI Coding, Model Cost https://www.winzheng.com/en/article/reflection-beam-501b-open-weight-moe-model-analysis Winzheng en 2026-10-17T06:11:15+08:00 Reflection Open-Sources 501B Model Beam: Only 23B Parameters Activated per Token, Claims to Match GLM-5.2 with a Quarter of the Compute 开源大模型, Beam, Reflection AI, 混合专家模型, MoE, GLM-5.2, 模型评测, AI竞争格局 https://www.winzheng.com/en/article/gpt-6-astra-pro-multi-agent-parallel-upgrade Winzheng en 2026-10-17T06:06:52+08:00 GPT-6 Astra Pro Adds Multi-Agent Parallelism: The Speedup Promise Is Met, but Quota and Control Questions Remain Open GPT-6, OpenAI, 多智能体, ChatGPT, AI智能体, Astra Pro