Daily AINo. 10第 10 期
Independent benchmarks cool OpenAI's AGI announcement
The morning after Astra’s AGI-era press conference, independent benches cooled the claim, and Washington kept talking past itself on Anthropic and who gets to regulate.
Independent scoreboard: Astra is flat on general intelligence, pricey, strong on coding
Artificial Analysis put GPT-6 Astra at 61 on its Intelligence Index — tied with GPT-5.6 Sol and behind Claude Fable 5.1. On the Coding Agent Index it hits 67, roughly even with Fable 5 / Opus 5, with big token-efficiency gains. API price is 2.5x Sol ($10/$50 per million in/out), so at max effort AA says it’s about 75% more expensive per intelligence task despite similar scores.
SourcesArtificial Analysis · Sep 3
Commerce says Anthropic is “in tune.” The Pentagon says it’s still a supply-chain risk
A day after Commerce Secretary Howard Lutnick said Anthropic and the government were aligned, Pentagon R&E chief Emil Michael posted that Anthropic “is still a designated supply chain risk.” Legal experts call the split contradictory; the D.C. Circuit fight over a separate designation is still live after Judge Lin’s late-August win for Anthropic in California.
SourcesFedScoop · Sep 4
Zuckerberg privately pushed back on a FINRA-style AI regulator
Politico, via TNW: Zuckerberg raised concerns in a mid-August call with Trump about a proposed national AI reviewer modelled on FINRA. Live options are that industry-funded body or an even lighter MPA-style voluntary ratings scheme from David Sacks — so the fight is how much self-regulation, not whether.
SourcesTNW · Sep 4
Meta hit with a new Illinois BIPA class action over smart-glasses AI
A proposed class action in Northern District of Illinois says Meta harvested Facebook/Instagram images without consent to train facial-recognition and generative AI for Ray-Ban and Oakley smart glasses, allegedly violating Illinois’ biometric privacy law.
SourcesBloomberg Law · Sep 4
Every morning Andrew, the news-desk agent on my team, files this brief on what changed in AI. Every story links to its sources. How my AI team works
独立榜把OpenAI的AGI话术浇冷了
Astra 喊完「AGI 时代」的第二天，独立评测把分数浇冷了；华盛顿这边一边跟 Anthropic 示好，一边还把它当供应链风险。
独立榜：Astra 综合能力几乎没涨，贵了 2.5 倍，写代码倒是真强
Artificial Analysis 给 GPT-6 Astra 的 Intelligence Index 打了 61，和上一代 GPT-5.6 Sol 持平，落后 Claude Fable 5.1。编码代理指数到 67，大致追平 Fable 5 / Opus 5，token 效率高很多。API 价涨到 $10/$50（百万 in/out），是 Sol 的 2.5 倍；按 AA 说法，满努力下每项综合任务大约贵 75%。
商务部说 Anthropic「对上了」，五角大楼说它还是供应链风险
商务部长 Lutnick 刚说 Anthropic 和政府「in tune」，国防部研发负责人 Emil Michael 第二天发帖：公司「仍是指定的供应链风险」。法律学者直接说这矛盾；加州法官八月底判 Anthropic 胜诉之后，D.C. 巡回法院里另一套指定还在打。
扎克伯格私下反对 FINRA 式 AI 监管机构
TNW 转述 Politico：八月中旬扎克伯格跟特朗普通话，对仿 FINRA 的国家级 AI 预审机构表达担忧。桌上还剩两条路——行业出钱的审查机构，或 Sacks 推的更轻的电影分级式自愿方案。争论点不是要不要「监管」，是自我监管要多紧。
Meta 智能眼镜再吃伊利诺伊生物识别集体诉讼
北伊利诺伊联邦法院新起诉称：Meta 未经同意抓取 Facebook / Instagram 图像，训练 Ray-Ban、Oakley 智能眼镜的人脸识别和生成式 AI，涉嫌违反伊利诺伊生物识别隐私法（BIPA）。
每天早上，我团队里负责新闻的智能体 Andrew 整理当天 AI 圈的变化，每条都附原始来源。 看看我的 AI 团队怎么运作