2026-09-20
AI 快讯日报.

百度搭子企业版升级并发布开发平台，亮相超级智能体大会深圳站 ↗

Qwen 开源 Qwen-Image-2.1：7B 统一生成与编辑并原生支持透明图像 ↗

Qwen offers comprehensive functionality spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts.

农业农村部：大力推进"人工智能 +"农业，拓展无人机、物联网等应用场景 ↗

农业农村部党组会议强调，要把培育壮大农机装备产业作为农业现代化的关键支撑，聚焦高端智能、丘陵山区适用农机等短板，攻关突破标志性整机和关键共性技术，并大力推进“人工智能 +”农业，拓展无人机、物联网等应用场景。#人工智能 + 农业#

Scaleout Systems 展示断网可用的机载 AI 无人机目标识别与打击方案 ↗

斯凯奥特系统放弃前沿大模型，转而用轻量化模型让无人机在前沿独立完成目标识别与打击。其联邦学习架构可在通信中断时继续迭代，

AI 与知识共享的毁灭：LLM 正在摧毁开源生态吗？ ↗

The balance of software copyright protection and openness has always been fraught with minutiae and detail that bores all but the most nerdy of pedants. Yet, through much effort and 40 years of debate we had reached an equilibrium. Now AI has thrown that out the window.

Epoch AI 调查：美国成年人近每日使用 AI 的比例半年内翻倍至 19% ↗

Surveys by Epoch AI and Ipsos suggest AI is becoming part of everyday life for more Americans. The share of US adults using AI almost daily more than doubled between March and August 2026, rising from 8 percent to 19 percent. Meanwhile, the share using it just once a week fell from 17 percent to 10 percent.

A20 Pro 芯片端侧 AI 实测：iPhone 18 Pro 本地跑 270 亿参数模型，速度较 iPhone 17 Pro 翻倍 ↗

iPhone 18 Pro 系列配备 12GB 96 位 LPDDR5X 内存，内存带宽达 115.2GB/s，双 16 核神经网络引擎峰值吞吐超 7 核 GPU。网友实测 Bonsai‑27B 模型运行流畅，但 Bonsai 2 因 2 比特量化体积过大无法完整载入本地内存。

微软与伊利诺伊大学提出 StudentSim：用模拟学生帮 AI 导师更快学习 ↗

Microsoft and the University of Illinois built StudentSim to replicate individual students from limited data and give AI tutors fast, low-cost feedback. In tests covering 60 students across chess, English, and math, it outperformed GPT-5.4. A chess tutor trained with StudentSim also earned the highest expert ratings among three versions tested.

石化盈科×火山引擎：TRAE 嵌入研发全流程，数千人组织向 AI Native 转型 ↗

特朗普宣布组建 AI Force 并计划任命 AI czar，主张放松对 AI 发展的监管 ↗

Trump announces an "AI Force" modeled after the Space Force and plans to appoint an "AI czar" with a "high IQ." He claims AI could reach 25 percent of US economic output, rejects new regulation, and frames criticism of data centers as a left-wing attack.

Anthropic 与埃森哲各投 10 亿美元开展前沿模型独立安全评估 ↗

双方承诺五年各投至少 10 亿美元，由埃森哲旗下 Faculty 主导对 Anthropic 前沿模型开展独立评估与红蓝对抗测试。评估人员将进驻企业内部，访问权限与普通员工相当，可核实安全承诺、上报安全事件。

据报道 Anthropic 将 IPO 从 10 月推迟至 11 月，估值预期约 2 万亿美元 ↗

Anthropic is reportedly delaying its IPO from October to November 2026 to present strong Q3 results. Investors expect a roughly $2 trillion valuation, but rising infrastructure costs, including $1.25 billion a month for its SpaceX deal alone, and unresolved security risks complicate the listing.

特朗普在 Truth Social 和 X 发起为人工智能改名投票，超 18 万人参与，卓越智能暂以 40.4% 领先 ↗

特朗普称 AI 风险是民主党炮制的骗局，将扶持 AI 发展，还计划组建 AI 部队、任命 AI 事务总管，并发起 AI 改名投票，目前“卓越智能”得票领先。

谷歌高管称亚洲 AI 普及速度被低估，两年走完云计算十多年历程 ↗

谷歌亚太及日本地区 AI 业务总经理哈沙 · 孔杜里表示，亚洲 AI 普及速度远超外界预估，该地区或成谷歌云增长最快市场。不过亚洲企业采用 AI 时更谨慎，更注重安全。他还预计企业对 AI 成本的关注将上升。

腾讯混元联合清华北大发布 WebCraftBench，用软件测试方法评测 AI 网页生成 ↗

雷神首款 AI 移动工作站 AI Master M7000 明日开售：锐龙 AI Max+ 395 处理器，1.99 万元 ↗

新机搭载 AMD 锐龙 AI Max+ 395 处理器，配 64GB 内存与 768GB+85GB AI SSD，内置牧马人本地推理引擎，支持群联 aiDaptiv+ 方案，可端侧运行 120B 大模型。

消息称 Anthropic 拟于 IPO 前推出新模型应对 OpenAI GPT-6 Astra 竞争 ↗

消息称 Anthropic 正评估下一代模型的安全水平，内部也在权衡研发投入与盈利节奏。GPT-6 Astra 已占企业 AI 支出约 13%，OpenAI 在 OpenRouter 上花费反超。

工信部副部长辛国斌：大力发展开源基座模型和垂类模型，深入推进 AI+ 制造 ↗

辛国斌在 2026 世界制造业大会提出四点主张，其中明确要深入推进“人工智能 + 制造”，梯度培育智能工厂，并加快开源基座模型与垂类模型发展。#2026世界制造业大会# #人工智能 + 制造#

阿里通义Qwen团队发布Qwen3.8-LiveTranslate，实时翻译平均延迟降至2.3秒、支持60种语言 ↗

Qwen has released Qwen3.8-LiveTranslate, its next-generation real-time simultaneous interpretation model. It listens to live speech, with optional video frames, and returns translated text and speech while the speaker is still talking. The core change is a new Interleave architecture. Qwen reports gains in faithfulness, fluency, and conciseness, with average lagging (LAAL) dropping from 2.8 seconds to 2.3 seconds. The release also adds real-time speaker diarization, synchronized bilingual display, and long-context disambiguation.

B站上线"AI 无限竞技场"大模型测评榜，GPT-6 Astra 居榜首 ↗

首轮榜单显示 GPT-6 Astra 暂居榜首，前五名中国产大模型占三席；测评由 UP 主自主命题，还原真实场景表现，排名实时更新