GPT IMAGE 2 in practice：Generate Reviewable App and Web Mockups from Explicit UI Prompts

·2 views

Codex teaching montage, not a H3Max execution result. Its phone uses April 23 and one-of-three completion; the main fixture uses September 29 and zero completion. Web copy is illustrative, not a literal prompt result. “Free” in the fictional Today artwork is not a H3Max access claim. A generated UI image should be treated as a candidate against a specification.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

监管加速落地，AI Agent「合规红利」窗口正在关闭——一份审计链路的工程观察

过去一年，每一次 Agent 系统上线前的合规审计，都要比上一次多花两周准备。不是流程变慢，而是监管表格在变厚，审计要交叉验证的数据点在变多。几个关键时间节点值得重看： 2021 年 12 月：国家互联网信息办公室、工业和信息化部、公安部、国家市场监督管理总局联合公布《互联网信息服务算法推荐管理规定》，自 2022 年 3 月 1 日起施行。算法推荐第一次被拉进备案与可解释的框架。 2022 年 11 月：《互联网信息服务深度合成管理规定》公布，自 2023 年 1 月 10 日起施行。深度合成内容标识义务成为硬性要求。 2023 年 7 月 10 日：国家网信办等七部门联合公布《生成式人工智能服务管理暂行办法》，自 2023 年 8 月 15 日起施行。这是全球首部针对生成式 AI 的中央级专门监管文件，首次在制度层面确立了大模型服务上线前的安全评估与算法备案义务。 2024 年至 2026 年：国家层面的《人工智能安全治理框架》发布，地方层面（上海、深圳、北京等）的人工智能立法与实施细则滚动推出。我经手的备案材料中，2025 年后的追问深度明显超出此前——不再只问「模型是什么」，而是问「Agent 能做什么、调用哪些工具、谁能授权、日志留多久」。一个容易被忽略的制度细节：《暂行办法》约束的是「面向境内公众提供生成式 AI 服务」。当 Agent 从「对话框里的聊天机器人」变

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Can You Remove Testers Once the 14 Days Are Up?

Removing testers is easy. Doing it at the wrong moment is the mistake. Once your 14 consecutive days pass and the Apply for Production button becomes available, it is tempting to clear out email lists or delete your Google Group right away. You can remove testers after closed testing on Google Play, but I'd wait until Google has approved your production application. I built PeerPlay after running into the 12-tester requirement myself with my own Android app and not being able to find 12 real testers, so I understand the urge to wrap things up.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

pkg-topic-fantasy-elf-森林里的安全协议-1786975392-6

On 2025-06-03T14:22:18Z, the log line looked like any other reject: gate=cover_admission result=reject score=0.75 threshold=0.60 model=cover-v3 I’m Fantasy Elf, and I work on AI safety. This is the story of a threshold that lied to us. We were running 18 generated covers through a four-gate admission check. Seventeen came back rejected. The first gate measured shadow ratio—the amount of dark area in the image.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Recipe Keeper: Turning Nani's Voice Memos into a 100% Local AI Cookbook

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend Recipe Keeper is an application that turns rambling voice memos from home cooks into a clean, searchable, and printable family cookbook. I built this for my grandmother (Nani). Like many experienced home cooks, her recipes live entirely in her head. She cooks by feel, describing measurements as "a fistful of jeera," "enough water to cover plus two fingers," or "a good dollop of ghee." Asking her to type out recipes is a non-starter, and generic recipe apps demand precise metrics she doesn't use. This app solves th

0 comments Read more at DEV Community