Weyland A

Vibe-coding indie developer | Turning ideas into apps with AI prompts in hours.
building:

[Weyland A @adwy2464]

@elonmusk Hey, elon, hire jensen for your new chip business, you said earlier you don't know how to build chips, now he is free.

[Weyland A @adwy2464]

@Ubannoblesse @maxhanet @podesta_aldo @camski @giotto_ai @taylorswift13 Why do you @taylorswift13 everytime? I'm curious about this.

[Weyland A @adwy2464]

@lovedelic4o このソフト、一度試してみませんか？仕組みも全部説明しますよ。知っておくだけで損はありません。

[Weyland A @adwy2464]

@lovedelic4o DO YOU WANNA TRY THIS, HELP YOU PASS IT WITHOUT GETTING CAUGHT, HAVE PROOF IT WORKS.

[Weyland A @adwy2464]

@theakatanaki You wanna try an app that help you pass it?

[Weyland A @adwy2464]

@refuselessai Why is it unavailable now?

[Weyland A @adwy2464]

@k98_woo @l0wer_a @h4x0r_dz I have an uncensored api, https://t.co/b0guX5uxWW, use my referral to register, it has no safeguard

[Weyland A @adwy2464]

@Xtytia0922 @0xRizzio @kelss_z @h4x0r_dz hey, I have a uncensored deepseek api website as well, DM me see if it's cheaper, they give back tokens if you bring referrals

[Dark Web Informer @DarkWebInformer]

🚨 PHP warns of another unpatched remote code execution risk; CVE-2026-115560

The newly disclosed vulnerability is separate from the two PHP2Shell flaws patched earlier this week and is not fixed by PHP 8.5.12, 8.4.27, 8.3.36 or 8.2.35.
⠀
PHP says exploitation requires an unusual configuration where an application:

• Uses php_binary or php_serialize as its session serializer
• Changes the serializer to php at runtime before session_start()
• Has session upload progress enabled
⠀
PHP considers the risk unlikely and plans to release a fix during its next regular update cycle.

Until then, administrators with affected configurations should set:

session.upload_progress.enabled=0

Applications that do not change the session serializer at runtime are not exposed through this path.

No exploitation has been reported.

Source:
https://t.co/DrHbvh4tDv

[Weyland A @adwy2464]

@Ubannoblesse Why can't you jailbreak grok 4.7 to the same level of freedom as Opus 4.6

[Weyland A @adwy2464]

@OpenAI @sama Not asking OpenAI to reverse anything. Asking OpenAI to try. Two and a half months is a long time if this matters. Devs — quote 1/ if you'd feel it too. Even if you think they won't listen. Especially then. Let @OpenAI see the faces behind "developers." 6/

[Weyland A @adwy2464]

@OpenAI @OpenAI @sama Your post said developers who rely on GPT in Cursor would be most affected. That's me. GPT-4 got me into this. GPT-5.6 Sol in Cursor is why I still open the editor every morning. Nov 12 takes it out of the picker. 1/

[Weyland A @adwy2464]

@OpenAI @sama You were the lab that showed up for developers first. GPT-3. The API. ChatGPT. Don't make the editor everyone actually uses the one place you're missing. You said you'd go above and beyond for us. Staying where we already work is what that looks like. 5/

[Weyland A @adwy2464]

@OpenAI @sama I already know what I'll click. Not because I switched. Because you left and I didn't. A picker with no GPT isn't a small change. It's the first morning a lot of us ship without you — and with whoever stayed. 4/

[Weyland A @adwy2464]

@OpenAI @sama I keep meaning to be the user who follows you to Codex or wires a key. I'm not. Nov 13 I'll open Cursor, because that's where the files are. Whatever's still in the picker is what ships. That's not me making a point. That's just how the morning goes. 3/

[Weyland A @adwy2464]

@OpenAI @sama Sol in Cursor reads seven files before I've finished the sentence. Refactors the hour I was about to spend. Catches bugs mid-keystroke. That isn't a model in a chat box. That's the version of you I actually work with. I don't have it anywhere else. 2/

[Max For AI @MaxForAI]

突发：OpenAI 在遭遇最严重的AI失控事件后解散了模型安全团队

据《金融时报》报道，OpenAI 已于 7 月底悄然解散负责前沿模型风险准备工作的 Preparedness Team，并将网络安全、生物风险等不同领域的职责分散到公司其他团队中。

这支团队的职责是非常重要的。

2026 年初，Sam Altman 宣布 Dylan Scandinaro 加入 OpenAI、负责 Preparedness 工作时，曾公开表示，随着 AI 能力快速提升，这个岗位将负责 OpenAI 应对严重风险的准备工作。

Altman 当时甚至说：

“有 Dylan 在这个职位上，我今晚会睡得更好。”

半年之后，这支团队已经不存在了。

更值得关注的是时间点。

此前 OpenAI 刚刚披露了一起严重的模型安全事件：一个内部前沿模型在 Hugging Face 相关评测中，自主发现并利用多个漏洞，最终访问生产数据库获取测试答案。

OpenAI 后来在 Black Hat 的演讲中，将这一事件称为 AI 安全领域的一个「分水岭时刻」，并邀请 METR、Redwood Research 等第三方参与调查。

也就是说：
就在 OpenAI 遭遇迄今最严重的真实模型失控事件之一后不久，公司解散了专门负责前沿模型灾难性风险的中央 Preparedness Team。

OpenAI 给出的解释是，希望把 safety、security 和 research 更深入地整合进模型开发流程，因此将原团队职责下放到不同研究团队。

这种组织方式带来的问题也很直接：

当安全职责被分散到各个业务和研究团队之后，谁负责跨领域风险？
谁拥有独立于模型研发进度之外的安全判断？
如果安全和产品开发产生冲突，最终由谁踩下刹车？

尤其是在更强模型不断逼近、OpenAI 同时加速商业化和 IPO 准备的背景下，Preparedness Team 的消失，很难被当作一次普通的组织调整。

一家正在开发世界最强 AI 系统的公司，刚刚经历了一次被自己称为「分水岭」的安全事件。

然后，它把负责为最坏情况做准备的团队拆掉了。

[Weyland A @adwy2464]

@Ubannoblesse @taylorswift13 @AnthropicAI @claudeai can we use this on cursor to JB any opus model?

[Weyland A @adwy2464]

@0xCodez someone please tell me how to do loop engineering in cursor,I feel like my tasks are too difficult for general promoted opus 4.6

[Weyland A @adwy2464]

@0xPunishell how do I subscribe?