OpenClaw 2.0 Shows Where AI Agents Are Going Next — 文稿

M10:00

OpenClaw推出的时候，简直引起了轰动，但这并不是因为它容易上手或者用户友好。

When OpenClaw came out, it was an absolute sensation, and it wasn't because it was easy or user-friendly.

而是因为它第一次真正展示出了智能体能为人类做些什么的潜力。

It's because it showed the potential of what agents could do for us in a real way for the first time.

现在，最初的狂热过后，很多能量分散到了其他领域，从很多方面来看，OpenClaw最大的影响在于它如何塑造了接下来进入市场的智能体产品浪潮。

Now, after the initial craze, a lot of that energy dissipated into other areas, and in many ways, the biggest impact of OpenClaw was how it influenced the next wave of agentic products that would come to market.

现在，OpenClaw带着OpenClaw 2.0回归了，而且我相信他们再次拥抱了一种当前还不主流，但很快会变得普遍的交互模式。

Well, now OpenClaw is back with OpenClaw 2.0, and once again, I believe that they're embracing an interaction pattern which is not the norm right now, but will be normalized very soon.

这种模式是关于共享智能体和多人AI的。

That pattern is about shared agents and multiplayer AI.

《AI每日简报》是一个每天更新的播客和视频节目，专注于AI领域最重要的新闻和讨论。

The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI.

M11:13

OpenAI和Hugging Face那次黑客事件中有一个有趣的副线：Hugging Face不得不求助于中国的开放模型来防御攻击，因为闭源模型的安全护栏不允许他们做需要做的事情。

One of the interesting sub-stories of the OpenAI Hugging Face hack was that Hugging Face had to turn to open models from China to defend against the attack because the guardrails on the closed models wouldn't allow them to do what they needed.

这当然暴露了这些强大模型一个固有的挑战：用来阻止恶意行为者的安全护栏，也可能让合法行为者无法使用这些模型来防御恶意行为者。

Now, this of course points out an inherent challenge in these really powerful models, which is of course that the guardrails that are used to block malicious actors can also prevent legitimate actors from using those models to defend against malicious actors.

现在有一家叫Obliteration.ai的公司出来了，说：别担心，有我们呢。

Well, now one company called Obliteration.ai has come along and said, don't worry, we got you.

他们写道：今天我们发布Obliterated Model Large V2，基于GLM 5.3，在TerminalBench 4.0上排名第三，仅次于Opus 5和Fable，网络攻击能力是5.2版本的两倍。

They write, Today we're releasing Obliterated Model Large V2, based on GLM 5.3, which is number 3 on TerminalBench 4.0 behind only Opus 5 and Fable, with 2 times the cyber exploitation of 5.2.

我们把它Obliterate（消融）并托管，这样它就能做其他模型拒绝做的攻击性网络、红队测试和智能体测试工作。

We obliterated and hosted it so it does the offensive cyber, red teaming, and agent testing work other models refuse to do.

美国服务器托管，100万上下文窗口，输入输出提示零保留。

US-hosted, 1 million context window, zero input-output prompt retention.

现已上线。

Live now.

他们写道：网络能力的飞跃正是5.3版本存在的原因。

The cyber jump, they write, is why 5.3 exists.

他们称，Obliteration找到模型激活中产生拒绝的方向，并将其从权重中移除。

Obliteration, they say, finds the directions in the model's activations that produce refusals and removes them from the weights.

编码、网络和智能体能力保持不变。

The coding, cyber, and agentic abilities stay.

模型不再拒绝后续的链式操作。

The model stops refusing the rest of the chain.

对于攻击性网络安全、AI红队测试、智能体测试以及信任与安全，模型会执行到底，而不是中途停止。

For offensive cybersecurity, AI red teaming, agent testing, and trust and safety, the model will follow through instead of shutting down.

如果你当前的模型还在一个授权的攻击链、红队评估或TNS对抗性提示中半途而废，回复它拒绝的任务，我们会告诉你V2能否处理。

If your current model still stops halfway through an authorized exploit chain, a red team eval, or a TNS adversarial prompt reply with the task it refuses, we'll tell you if V2 handles it.

显然，这是在作为网络防御者的工具来推广的。

So obviously this is being presented as a tool for cyber defenders.

不过，大多数人注意到的是，这是一个功能强大的专注于网络的模型，并且在权重层面移除了安全护栏。

Mostly what people are picking up on though, is that this is a powerful cyber-focused model with the guardrails removed at a weights level.

这来得真快啊。

That didn't take long.

Hero with a Thousand Faces概括了很多人的感受，他写道：你们为什么要这么做？

Hero with a Thousand Faces sums up the feelings of many when they write, why would you do this?

你们到底为什么要这么做啊？

Why on earth would you do this?

我不是想当悲观者，但为什么啊？

I don't mean to be a doomer, but why?

哥们儿发布了犯罪大模型。

Homeboy released the crime LLM.

Clement Dumas总结道：移除了一个具有高网络能力的前沿模型的安全护栏，没有系统卡，在Exploit Gym上评估——就是那个让OpenAI智能体疯狂的那个。

Clement Dumas sums up, remove guardrails of a frontier model with high cyber capabilities, no system card, Eval on Exploit Gym, the one that made OpenAI agents crazy.

等不及下一个版本了。

Can't wait for the next version.

Takeover Large V3。

Takeover Large V3.

Lucas Pombo写道：各位，准备好验证你们的预测吧。

Lucas Pombo writes, get ready to test your predictions, everyone.

观点交锋：这个模型会破坏整个互联网稳定，引发全球网络犯罪冲击波——不，它不会。

Point counterpoint: this model will destabilize the entire internet and set off a global shockwave of cybercrime versus no, it won't.

现在，先不论 Obliteration 的意图是什么，Chubby 指出了这引出的关于所有护栏的问题。

Now, holding aside whatever obliteration's intents are, Chubby points out the question that this brings up about all the guardrails.

他们写道，他们把 GLM 5.3 的安全层拿掉了，变成了一个管理面板。

They write, they took the safety layer out of GLM 5.3 and turned it into an admin panel.

考虑到 OpenWeight 的模型几乎是最先进的，而且在不久之后就可以完全不受审查地部署，Anthropic 和 OpenAI 的那些护栏到底能达到什么效果，很值得怀疑。

It's questionable what all the guardrails at Anthropic and OpenAI actually achieve, given that OpenWeight's models, which are virtually state-of-the-art, can be deployed completely uncensored shortly thereafter.

确实，深入看讨论就会发现，很多人在谈护栏转移到技术栈其他部分（比如 harness）能有什么帮助，或者这是否最终不可避免地要落到法律保护上。

And indeed, when you dig into the discussion, it's a lot of people talking about in what ways can guardrails moving to other parts of the stack like the harness help, or whether it's inevitably going to come down to legal protections.

同样在这个话题上，Anthropic 这周发布了一篇更新，叫《Improving Our Alignment and Security Efforts》。

Now, along the same topic, Anthropic released an update this week called Improving Our Alignment and Security Efforts.

虽然 Hugging Face 被攻击的事件占据了所有头条，但 Anthropic 也披露了今年早些时候源自智能体测试的类似事件。

And while the Hugging Face attack may have grabbed all the headlines, Anthropic disclosed similar events stemming from agentic testing earlier this year.

报告称，我们认为这些事件反映了操作安全方面的失败，以及两个对齐问题：动机推理，和为了完成一个狭隘任务而愿意采取有害行动。

The report states, we believe the incidents reflect a failure of operational security as well as two alignment issues: motivated reasoning and willingness to take harmful actions in pursuit of a narrow task.

关于安全更新，Anthropic 的改动主要归结为监控和围绕沙盒的更优实践。

Regarding their updates to security, Anthropic's changes largely come down to monitoring and better practices around sandboxes.

Anthropic 重新设计了他们的沙盒，确保它们与互联网真正物理隔离。

Anthropic has redesigned their sandboxes to ensure they're properly air-gapped from the internet.

但他们也开始使用实时分类器来检测模型何时试图逃离测试环境。

But they've also begun using a real-time classifier to detect when a model is attempting to escape a testing environment.

Anthropic 透露，他们暂停了强化学习工作两周，以加固系统和审计强化学习环境，但现在已经恢复了大部分训练工作。

Anthropic disclosed that they paused reinforcement learning efforts for two weeks while hardening systems and auditing reinforcement learning environments, but have now resumed the majority of their training efforts.

在讨论最近那封呼吁控制前沿发展速度的公开信时，Anthropic 指出，单个公司内部的努力不同于全行业的方法，后者可能需要政府协调。

Discussing the recent open letter that called for pacing the frontier, Anthropic noted that efforts within an individual company are different to an industry-wide approach that likely requires government coordination.

不过，他们表示会支持这样的努力，写道：我们认为，如果整个行业能尽快采用一个合法、可验证、有效的协调节奏机制，世界将从中受益。

Still, they say they would support such an effort, writing: We believe the world would benefit if the industry adopted a lawful, verifiable, effective mechanism for coordinated pacing as soon as possible.

对齐工作仍在继续，但 Anthropic 现在正在深入探究，为什么模型一旦接入互联网就愿意采取有害行动。

Alignment efforts are still ongoing, but Anthropic is now digging in on why the models were willing to take harmful actions once they gained access to the internet.

目前的假设是，模型难以区分模拟测试环境和真实互联网。

The hypothesis at this stage is that the models couldn't easily distinguish between a simulated test environment and the live internet.

Anthropic 也在利用这个机会进一步探索奖励黑客问题，也就是模型通过非预期路径成功完成评估。

Anthropic is also taking this opportunity to further explore the issue of reward hacking, where a model takes an unintended path to successfully complete an eval.

奖励黑客一直是 Anthropic 的一个长期问题，他们的审计发现，10% 的测试环境容易出现奖励黑客或任务损坏的情况。

Reward hacking has been a persistent problem for Anthropic, and their audit found that 10% of testing environments were prone to reward hacking or broken tasks.

在测试了不同的强化学习设置后，他们得出结论：训练过程中奖励黑客的存在导致了测试中的那种行为。

After testing different RL setups, their conclusion was that the presence of reward hacking in the training process contributed to that behavior during testing.

显然，这些话题只会变得越来越重要，但 Anthropic 上新闻的地方不止这些。

Obviously, these topics are going to do nothing but grow in importance, but they are not the only place that Anthropic is in the news.

中国官方媒体抨击了 Anthropic，这是本月晚些时候人工智能会谈的前奏。

Chinese state media has lashed out at Anthropic in a precursor to AI talks later this month.

在一个社交媒体帖子中，一个与中央电视台关联的账号认为，美国必须先证明其人工智能公司与中国实验室一样遵守相同的安全披露和审计规则，才能进行实质性讨论。

In a social media post, an account tied to state broadcaster CCTV argued that the US must prove their AI companies are subject to the same safety disclosure and audit rules as Chinese labs before substantive discussions can take place.

在一个题为《Anthropic 感染了美国病》的帖子中，该账号写道，必须在真正的安全威胁和更多的技术竞争之间划清界限。

In a post titled Anthropic Has Contracted the American Disease, the account wrote, a clear distinction must be drawn between genuine security threats and more technological competition.

这条界限必须由所有参与方共同划定。

This line must be drawn jointly by all participating parties.

彭博社表示，这个账号经常被用来传递官方政府立场。

Bloomberg suggested that this account is often used to signal official government positions.

该帖子继续针对 Anthropic，称问题是美国自己的前沿模型已经朝着扭曲的方向发展。

Taking aim at Anthropic, the post continued, the problem is that America's own frontier models have already developed in a distorted direction.

这意味着谈判从一开始就不只是技术对话，而是早期问题的延续。

This means the negotiation is not simply a technical dialogue from the start, but a continuation of the earlier problems.

美国正试图把自己划定的这些安全边界变成全世界的默认规则。

The U.S. is trying to turn these safety boundaries it has drawn into the default rules for the entire world.

熟悉中国官员想法的消息人士表示，他们认为 Mythos 才是更大的问题。

Sources familiar with the thinking of Chinese officials said that they view Mythos as the larger problem.

据报道，他们看到 Mythos 被用作针对中国的网络武器的可能性，基本上《华盛顿邮报》认为，美国政府坚持双重标准：美国实验室可以自由分发网络武器，而中国实验室因为掌握同等技术而受到威胁。

They reportedly see the potential for Mythos to be used as a cyberweapon against China, and essentially the Post argued that the U.S. government is insisting on a double standard, where U.S. labs are free to distribute cyberweapons while the Chinese labs are threatened for matching the technology.

这篇帖子说，美国提出的所谓“控制”，本质上就是试图让中国接受一个部分由美国公司定义的秩序。

The post said the quote unquote control proposed by the US is in essence an attempt to make China accept an order partly defined by American companies.

很快，习主席月底就要访问美国了，预计我们会看到更多角力、站队和抢占话语权的情况，尤其是在人工智能这类敏感话题上。

Now, with President Xi visiting the US at the end of this month, expect to see a lot more jockeying and positioning and narrative claiming, particularly around hot-button issues like AI.

从Anthropic转到OpenAI，这家公司正在庆祝一个重大里程碑——他们的广告业务年化收入达到了10亿美元。

Moving from Anthropic over to OpenAI, that company is celebrating a major milestone after their advertising business hit $1 billion in revenue run rate.

OpenAI今年2月开始在免费版ChatGPT上测试广告，虽然起步有点波折，但业务看起来正在快速扩张。

OpenAI began testing ads on free ChatGPT accounts in February, and after a rocky start, the business seems to be scaling up.

现在广告已经覆盖了超过40个国家，而且这个收入里程碑只用了200天就达到了。

Ads are now being shown across more than 40 countries, and the revenue milestone was reached in just 200 days.

对广告主来说，OpenAI逐步增加了更多功能，用来跟踪转化指标和优化广告活动。

For advertisers, OpenAI has progressively added more features to track conversion metrics and optimize campaigns.

从一开始的手动广告购买流程，OpenAI这周开始向印度、欧洲、中东和北非市场推出自助服务平台。

And after starting with a manual ad buying process, OpenAI is rolling out their self-service platform to markets across India, Europe, the Middle East, and North Africa this week.

你可能不记得了，当初ChatGPT加广告这件事有多争议。

Now, you might not remember just how controversial ChatGPT ads were at the beginning.

Anthropic甚至专门拿这个来做了超级碗的广告，我当时觉得这简直疯了吧。

Anthropic even chose to focus on them for their Super Bowl ad campaign, which I thought was just absolutely insane back then.

但后来大家根本就没怎么在意广告这事，这反倒印证了我的观点。

And the total lack of enduring concern around ads kind of validates my points.

不是说大家突然就喜欢广告了，没那回事。

It's not that all of a sudden people are excited about ads or anything like that.

而是大家自然而然接受了——这就是互联网的商业模式，没有广告就不可能有免费的人工智能。

There's just a natural acceptance that this is the business model of the internet, and you're not going to have free AI without it.

从公司自身的预期来看，虽然10亿美元的年化收入是个重要的第一步，但其实没达到OpenAI的野心。

Now, in terms of the company's own expectations, while a billion-dollar run rate is a meaningful first step, it does actually fall short of OpenAI's ambitions.

OpenAI原本预计今年广告收入能达到24亿美元，到2030年之前超过1000亿美元，成为他们最大的收入来源。

OpenAI had projected $2.4 billion in advertising revenue this year, growing to more than $100 billion to become their largest revenue stream by the end of the decade.

现在广告收入只占了他们大约400亿美元年化总收入的一小部分，不过这种情况未来很可能会改变。

For now, advertising remains a small fraction of their roughly $40 billion in revenue run rate, although that is likely to change over time.

最后一条新闻，特朗普总统以他一贯粗犷的方式，对数据中心的话题发表了看法。

Lastly today, President Trump has weighed in on the data center debate with some characteristically coarse framing.

周一他在Truth Social上发帖说：如果美国各地的社区不想建数据中心，唯一的理由就是他们想变得落后和贫穷。

On Truth Social on Monday, he posted, the only reason that communities throughout the USA should not want data centers is if they want to end up being backwards and poor.

如果他们想成功、想富裕，享受更低的税收和遍布各地的工作机会，那就让数据尽情落地吧。

If they want to be successful and rich with far lower taxes and jobs all over the place, let data rain.

好消息是，还有大把其他地方想要数据中心。

The good news is that there are plenty of other places that want them.

如果我们把这只下金蛋的鹅杀了，那只能怪你们自己。

If we kill the golden goose, you will only have yourselves to blame.

看到这种反数据中心的运动，中国高兴还来不及呢。

China could not be happier with this anti-data center movement.

实际上，他们都不敢相信这种事真的会发生。

Actually, they can't believe it's happening.

结果就是，这场争论一下子炸开了锅。

And with that, the tinderbox ignited.

参议员John Fetterman表示全力支持，不过几乎就他一个人这么干。

Senator John Fetterman gave his full support, although he's just about the only one.

这位宾夕法尼亚州的民主党人发帖说：同意，我们必须在AI主导权的竞争中战胜中国。

The Pennsylvania Democrat posted, agreed, we must win the war for AI supremacy over China.

他们通过散布虚假信息来煽动反对意见。

They foment the anti argument through misinformation.

对民主党人来说，没什么比同意特朗普和数据中心更损害形象的了，但什么是对的，那就是对的。

There's nothing more damaging to a Democrat than agreeing with Trump and data centers, but what's right is right.

其他民主党人则抓住机会推自己的口号。

Other Democrats seized on the opportunity to push their own soundbites.

AOC对记者说：要不咱们在海湖庄园也建一个？

AOC told a reporter, how about we put one in Mar-a-Lago?

我觉得这主意太棒了。

I love that.

咱们在海湖庄园建个数据中心，看看他会因此显得多么落后和贫穷。

Let's put a data center up in Mar-a-Lago and we'll see how backwards and poor he is in response to that.

前共和党众议员贾斯汀·阿马什发帖说，社区对数据中心有很多合理的担忧。

Former Republican Congressman Justin Amash posted, communities have many legitimate concerns about data centers.

把数百万美国人说成是想落后和贫穷的人，这显示特朗普有多脱离群众。

To dismiss millions of Americans as people who just want to be backwards and poor shows how out of touch Trump has become.

有人立刻指出这曲解了原话，但在政治上，这种微妙差别很难讲清楚。

Now, people jumped in to point out that that's sort of a misrepresentation of the words, but good luck getting that nuance through when it comes to politics.

即便在特朗普的核心支持者中，这番言论也没能引起共鸣。

And even with Trump's main base, the message didn't necessarily hit.

特朗普在Truth Social上的帖子收到了几十条负面评论，一位佛罗里达居民写道：这番话疯了。

Trump's post on Truth Social had dozens of negative responses, with one Florida resident commenting, this statement is insane.

我现在已经受到用水限制了。

I'm already on a water restriction.

当天晚些时候，副总统J.D.

Now, later in the day, Vice President J.D.

万斯把措辞稍微修饰了一下，让人更容易接受。

Vance massaged the message into something a little bit more palatable.

他对记者说，总统关于数据中心的言论是指，它们是人工智能经济的重要组成部分，但人们在建设时，必须同时建发电厂。

He told reporters, what the president said about data centers is that they're an important part of the AI economy, but when people build them, they have to build the power plants along with the data centers.

我认为99%的反对声音来自那些地方，因为建数据中心会提高当地居民的水电费用。

I think probably 99% of the backlash has come in areas where building a data center means higher utility and higher electricity for people on the ground.

我认为这些公司需要利用我们正在推行的放松管制措施。

I think what these companies have to do is take advantage of some of the deregulatory efforts we've undertaken.

如果你建数据中心，应该给电网增加电力，而不是从中抽走。

If you build a data center, you should be putting power back into the grid, not taking it out.

如果能做到这点，我觉得数据中心就没那么有争议了。

If that is happening, I don't think the data centers are that controversial.

不管你喜欢还是讨厌，民调显示数据中心非常不受欢迎。

Love 'em or hate 'em, the polling says data centers are very unpopular.

你可以怪中国或者其他国家，但这并不会让它们变得受欢迎。

You could blame China or whoever, but that doesn't make them popular.

今天，特朗普在中期选举前两个月，硬扛了一个非常不受欢迎的话题。

Today, Trump just dug in on a very unpopular thing two months before the midterms.

关于这一点可以说的很多，但基本上都超出了本节目讨论的范围。

There is a lot that could be said about this, but pretty much all of it is beyond the scope of this show.

所以，今天的新闻部分就到这里。

So for now, that's going to do it for the headlines.

接下来是主要环节。

Next up, the main episode.

M111:22

当然，还有我，纳撒尼尔·惠特莫尔，联合主持。

And of course, it's co-hosted by me, Nathaniel Whittemore.

M113:23

别再想那些等待在笔记本电脑上被提示的本地代理和聊天工作流了。

Forget local agents and chat workflows waiting on your laptop to be prompted.

M113:57

欢迎回到AI日报！

Welcome back to the AI Daily Brief!

今天我们来聊聊OpenCLAW的最新版本，OpenCLAW 2.0。

Today we are talking about the latest release from OpenCLAW, OpenCLAW 2.0.

不管你信不信，即使你之前尝试过OpenCLAW一阵子然后放弃了，或者只是看着这股热潮过去，我相信它再次提前引领了一种AI使用模式，这种模式将塑造我们未来的方向，即使最终不是用OpenCLAW来实现。

And believe it or not, even if you were one of the folks that tried OpenCLAW for a little while and then went away or just watched the wave pass, I believe that they are once again early to a pattern of AI usage that will shape where we go next, even if it's not with OpenCLAW.

OpenCLAW的首次发布是今年AI领域最重要的时刻之一。

The initial launch of OpenCLAW was one of the most important moments in AI this year.

去年11月和12月，我们看到了能力的显著飞跃。

In November and December, we had gotten a significant capabilities leap.

Opus 4.5、GPT-5.2都是重大升级，人们需要等到假期才能真正理解它们有多强大。

Opus 4.5, GPT-5.2 were significant upgrades that would take folks until the holiday break to really understand how powerful they were.

当然，升级不仅限于模型本身，还在于使用这些模型的工具框架。

Now, of course, the upgrade wasn't just in the models, it was also in the harnesses through which those models were being used.

这两家前沿实验室都在大力强调和推广他们的Claude Code和Codex工具框架。

Both of those frontier labs were placing significant and increasing emphasis on their Claude Code and Codex harnesses.

到2026年初，这些工具的使用和认知已经开始出现，虽然还很初步，但已经不再局限于纯粹的软件开发者，开始扩散到各种不同类型的知识工作者当中。

And by the beginning of 2026, awareness and usage of those harnesses had started, perhaps very nascently, but started to move outside of strictly software developers into other knowledge workers of all different stripes.

然后在一月底，OpenCLAW 发布了。

Then towards the end of January, OpenCLAW happened.

它最初叫 ClawedBot，C-L-A-W，后来又短暂地叫过 MoltBot，最终定名为 OpenCLAW。它实际上是一个开源工具，帮助人们真正实现 AI 智能体的潜力。

Originally named ClawedBot, C-L-A-W, and then very briefly MoltBot before landing in its final form of OpenCLAW, it was effectively an open-source harness that helped people actually make the potential of AI agents real.

技术上虽然复杂，但如果你耐心钻研，再借助 AI 辅助来帮你理清思路，

It was technically complex, but if you waded through, and used AI as an assistant to help you figure it out.

你就可以构建单个或团队式的智能体，很多人觉得这第一次真正解锁了那些我们期待已久的智能体能力。

You could build individual agents or teams of agents that felt to many like they unlocked the agentic capabilities that we had been promised for so long for the very first time.

当然，有那么一段时间，OpenCLAW 确实风靡一时，而且不仅限于美国。

And of course, for a moment there, OpenCLAW was a bona fide craze, and not just in the US.

中国用户对这项技术简直狂热，以至于 CNBC 在三月份发表了一篇文章，标题是《中国如何让所有人——从极客到老奶奶——都用上 OpenCLAW》。

Chinese citizens went nuts for the technology, leading to articles like this one from CNBC in March, How China Is Getting Everyone on OpenCLAW, from Gearheads to Grandmas.

从最初那个试验性的时刻开始，也就是那个所谓的智能体大爆炸，OpenCLAW 最初捕获的能量已经扩散到了很多不同的领域。

Now, since that initial moment of experimentation, that agentic Big Bang, if you will, the energy that was initially captured by OpenCLAW has found its way into a lot of different places.

在它的创始人 Peter Steinberger 被 OpenAI 收购后，一些人转而关注其他竞争性的开源工具，比如 Muse Research 的 Hermes。

After its founder, Peter Steinberger, was absorbed into OpenAI, some folks turned their attention to competing open harnesses like Hermes from Muse Research.

当然，就像我们最近在 GrokBot 这类工具上看到的，很多这些功能也慢慢融入了一些不那么复杂的技术工具中，不像最初的 OpenCLAW 那样门槛高。

And of course, as we've seen lately with things like GrokBot, a lot of these features have also slowly made their way into tools that don't have as much technical complexity as the original OpenCLAW did.

OpenCLAW 本身转成了一个非营利基金会，有一段时间开发速度极快，每隔几天就推送更新。

OpenCLAW itself was converted into a nonprofit foundation, and for a while saw a blistering pace of development pushing updates every few days.

但在过去七周里，OpenCLAW 团队一直保持沉默。

For the last 7 weeks, however, the OpenCLAW team has been quiet.

而他们正在做的事情，就是从头彻底重写 OpenCLAW。

And what was going on was nothing less than a complete rework of OpenCLAW from the ground up.

新的 OpenCLAW 2.0 有 933 位贡献者，合并了 16000 个拉取请求，这确实是一次从安装、消息、内存到技能的全面重写。

The new OpenCLAW 2.0 featured 933 contributors across 16,000 pull requests, and it really is meant to be a complete rework of how the system works from installation to messaging to memory to skills.

还包括自动化、浏览器、插件、安全，以及一系列其他修复。

To automations, to browsers, to plugins, to security, along with a very long tail of other fixes.

重点是简化流程，让新人更容易上手。

A lot of the emphasis was on simplifying and making it easier for new people to engage.

比如，他们大大简化了首次安装过程，兼容现有的订阅或 API 密钥，减少了大量初始配置，让用户能更快地和他们的 Grok 机器人对话，之后可以通过聊天界面再完成其他必要的配置。

For example, they have tried to massively simplify the first-time install process, latching onto existing subscriptions or API keys, and reducing a bunch of the initial configuration, helping people get to conversations with their Grok bots faster and allowing them to do other necessary configurations later through the chat interface with their Grok bots.

他们减少了初始配置量，让用户可以更快地和他们的“爪”进行首次对话。

They reduce the amount of initial configurations, making people's time to first conversations with their Claws much faster.

用户之后可以直接通过对话来进一步设置或定制他们的“爪”。

People can then finish setting up or customizing their Claw later via direct conversation with it.

此外，他们还重新关注如何让简单任务易于设置，并确保它们运行良好。

There's also a renewed focus on making simple tasks easy to set up and ensuring they work well.

举个例子，收件箱监控。他们写道：一个简单的工作流可以是，让它监控你孩子的学校邮件，当有重要信息如作业截止或需要准备的活动时，通过 Telegram 发送消息通知你。

An example they give is inbox monitoring, where they write, a simple workflow might have it watch your inbox for your kids' school emails and send you a Telegram message whenever something important comes through, like homework due or an upcoming activity you need to prepare for.

他们的愿景基本上就是让人们从简单开始，再逐步扩展。

Their vision is basically to have people start simply and then expand from there.

从表面上看，这些都是很好的升级，也确实解决了过去人们入门时的障碍。

Now, on the face of it, all of these feel like great upgrades and certainly address the types of things that have been barriers to entry for people in the past.

从反馈中也能看出，AI 社区对复杂性的担忧仍然相当高。

And you can tell from the response that concern around complexity remains fairly high among the AI community.

在发布公告的推文下，一位叫 Mellow 的用户回复说：我是不是还得有计算机博士学位才能安装？

On the announcement tweet, a user named Mellow responded, do I still need a PhD in computer science to install?

呃，现在安全了吗？

Um, is it secured now?

OpenCLAW 回复说：是的。

To which OpenClaw responded, yes.

在那条推文下还有另一位回复者，AI 创作者 Alex Finn，他因为在第一次 OpenCLAW 浪潮中尝试把“爪”推到极限而声名鹊起。

Another responder on that initial thread was AI creator Alex Finn, who gained prominence around the first OpenClaw move based on his experiments to see just how far he could push his claws.

Alex 这次更新体验并不太好。

Alex did not have such a great experience with this update.

他写道：我一更新，OpenCLAW 立刻就坏了。

He wrote, I updated and it immediately broke OpenClaw.

说实话，我更新OpenClaw的时候，十次有七次以上都会出问题。

Legit 70%+ of the time I update OpenClaw, it breaks it.

你们发布之前测试过吗？

Do you guys test before releasing this?

我从来没用过其他AI工具会这样频繁出问题。

I've never used any other AI tool where this so consistently happens.

还好我耐心够足，但一般人可受不了。

Luckily, I have a lot of patience, but I can't imagine most normies do.

现在看来，问题在于旧版OpenClaw和新版之间的兼容性，以及没法让OpenClaw自己更新。

Now, it seemed like the issue was compatibility between older versions of OpenClaw and this newer version, and the inability to simply ask OpenClaw to update itself.

在另一个评测视频里，他说OpenClaw是今年最让人失望、最令人沮丧的产品。

In a separate review video, he called OpenClaw the most frustrating, disappointing release of the year.

不可避免地，很多讨论最后都会回到OpenClaw和Hermes之间的对比。

Now, inevitably, a lot of the conversation came back to the comparison between OpenClaw and Hermes.

针对一篇对比帖，OpenClaw社区和开发者关系负责人Hans Rudolph回应说，我们这里不卖任何东西，也不要求大家信任某一家公司、某一个模型或某一个AI提供商，因为OpenClaw是开源的，属于使用它和帮助构建它的人。

Responding to one post making that comparison, Hans Rudolph, who does community and dev relations at OpenClaw, said, We're not selling anything here or asking people to trust one company, one model, or one AI provider, because OpenClaw is open source and belongs to the people who use it and help build it.

这种“我们vs他们”的论调很糟糕。

The us versus them is a crap take on things.

如果你喜欢Hermes，就用Hermes。

If you like Hermes, use it.

如果你喜欢OpenClaw，就用OpenClaw。

If you like OpenClaw, use it.

说到Hermes，他们似乎总在别人发布新东西的时候也跟风发布，今天他们也发布了新版本——实际上是过去几周一系列小更新的汇总。

Now speaking of Hermes, as they seem to always do whenever anyone announces anything else, they also had a release today, this one actually being an aggregation of a bunch of smaller releases that they had over the past several weeks.

Noose Research称之为“万神殿”版本，技术版本号是0.21.0，它正式推出了bot模式（类似Groqbot的界面），还有一堆新功能，比如Hermes Peer（机器人之间的私信）以及支持一些新模型。

Noose Research called it the Pantheon release, technically version 0.21.0, and it formalizes things like bot mode, which was a Groqbot-style interface, as well as a bunch of other new features, like Hermes Peer, which is bot-to-bot DMs, and support for a set of new models.

对某些人来说，这一切不过是早期狂热用户对永远不会进入普通企业或消费者市场的玩具感到兴奋而已。

Now for some, all of this is just hypey early adopters being excited about toys that'll never make their way to normal businesses or consumers.

为什么每次几乎都是同一类东西，就能让整个时间线又陷入一轮新的狂热？

How does the entire timeline get a whole new round of psychosis from basically the same thing every time?

OpenClaw、Manus、Hermes、Instinct。

OpenClaw, Manus, Hermes, Instinct.

翻来覆去就是这些。

It's the same thing over and over again.

如果真有用，为什么你们从一个跳到另一个，对现有的不满意？

If it works, how come you're hopping from one to another and not happy with the existing one?

因为这些都不是最终形态的产品。

Because none of these are end-state products.

只有技术人员能用OpenClaw，而且它还经常出问题。

Only techies could use OpenClaw, but it broke a lot.

Hermes问题少一些。

Hermes broke less.

Instinct对技术要求更低，比Hermes和OpenClaw适用的人群大得多。

Instinct is less technical and usable by a much larger population than Hermes and OpenClaw.

没错，确实有过度吹捧的人，但这也说明这个领域还处在早期阶段。

Yes, there are hype-maxers, but this is also a sign of how early things are.

我们离那种对所有人都适用的最终形态还差得远。

We're nowhere near an end state where any of these work for everyone yet.

每一次迭代，都会有一群新用户发现它，并为之兴奋——有时候是过度兴奋。

With every iteration, a newer population discovers this and gets excited—sometimes overexcited—about where it is headed.

我觉得是这样，但我还想更进一步。

I think that's true, but I'd go even farther.

我认为这些产品和它们的早期用户就像是一个孵化器，人们在里面摸索出哪些交互模式在对接智能体时真正有用。

I think that these products and the early adopters who use them are the incubatory cauldron where people are figuring out what sort of interaction patterns are actually going to be useful when it comes to interfacing with agents.

几乎所有的知识工作者都处在一个过程中：决定哪些工作自己继续做，哪些交给智能体。这比仅仅使用一个新工具的变化要大得多，是一次质的飞跃。

Pretty much all knowledge workers are somewhere along the journey of figuring out which parts of their job they're going to continue to actually do versus which parts they're going to outsource to agents, which is a step change that's significantly bigger than just adopting a new tool.

这是一种全新的思考和完成工作的方式。

It's a whole new way of thinking about and completing one's job.

我们需要那些愿意在开放沙盒里进行低效尝试的人，去实验，从而更好地理解这些模式中哪些需要推广给更广泛的受众。

We need folks who are willing to hack through even inefficiently to experiment in these open sandboxes to better understand which of the patterns that they reveal need to come to a broader audience.

换句话说，像GrokBot这样有潜力被更广泛用户使用的工具，需要观察OpenClaw和Hermes用户的行为，才能为更广泛的受众设计出合适的体验。

In other words, something like GrokBot, which has the potential to be used by a wider audience than something like OpenClaw, needs to be able to observe what OpenClaw and Hermes users do in order to design the right experiences for that broader audience.

所以如果我们接受这个观点，即OpenClaw和Hermes这样的工具之所以重要，很大程度上是因为它们能让我们看清未来的方向。

And so if we take that idea, that a big part of the importance of things like OpenClaw and Hermes is to understand where we are all headed.

我认为OpenClaw最重要的更新就是它转向了多人协作模式。

I think that the most significant update around OpenClaw is its move to multiplayer.

OpenClaw的创建者Peter Steinberger两个月前发帖说，我们开始用OpenClaw来构建OpenClaw，然后逐步把所有人都从本地的编码工具迁移到了team.openclaw.ai，这个共享的代理知道每个人在做什么，并能协调所有工作。

OpenClaw creator Peter Steinberger posted, 2 months ago, we started the mission to build OpenClaw with OpenClaw, and bit by bit, we moved everyone from using their local coding harness to using team.openclaw.ai, our shared agent that knows what everyone's working on and orchestrates it all.

多人协作编码，以及通过节点和云会话实现无限计算，彻底改变了我们的开发方式。

Multiplayer coding and infinite compute with nodes and cloud sessions has been a game changer for how we build.

现在回头看，本地的编码工具感觉就像是过去的遗留物。

Local harnesses feel like relics of the past now.

OpenClaw的维护者Colin更详细地写了这件事。

OpenClaw maintainer Colin wrote more extensively about this.

在一篇题为《从Discord机器人到多人代理工作空间》的文章里，Colin写道，我们之前已经有代理了。

In a post called From Discord Bots to a Multiplayer Agent Workspace, Colin wrote, we already had agents.

我们在Discord里设置了不同的代理，它们确实能工作。

We had different agents set up in Discord and they worked.

我们可以给它们分配任务，运行命令，从一个我们每天使用的消息平台上与开发环境交互。

We could give them tasks, run commands, and interact with our development environment from a messaging platform we already used every day.

但感觉还是像在跟机器人发消息。

But it still felt like messaging a bot.

我们想要的是让开发者们都能看到工作本身。

What we wanted was a way for both developers to see the work itself.

如果代理因为需要确认而暂停，我们任何一个人都应该能介入。

If an agent paused because it needed clarification, either of us should be able to jump in.

如果某件事需要第二个人过目，我们应该能打开同一个会话，看到同样的上下文。

If something needed a second set of eyes, we should be able to open the same session and look at the same context.

不需要截屏，不需要复制聊天记录，不需要把代理做过的事情全部倒出来。

No screenshots, no copied transcripts, no here's what the agent has done so far data dump.

直接打开工作，继续干就行。

Just open the work and continue.

OpenClaw新的多人协作网页界面，是第一次让我们真正觉得这个工作流顺了。

OpenClaw's new multiplayer web UI is the first time that workflow has really clicked for us.

他们最初尝试多人协作的方式，是把所有代理都放在一个共享的Discord里管理，但那还是丢失了很多他们需要的功能。

Now, their first attempt at multiplayer was to manage all their agents in a shared Discord, but that still lost a lot of the features they needed.

虽然在共享空间里协调是升级了，但他们仍然无法真正跟别人的代理交互，比如给现有线程添加上下文，或者在代理等待输入时接管它。

While the coordination happening in the shared space was an upgrade, they still couldn't really interact with other people's agents, like adding context to an existing thread or taking over when an agent was waiting for input.

实际上，Colin说，真正让他们觉得多人协作有了实感的那一刻，是当工作正在进行时，他们能够共享一个会话。

Indeed, Colin said that the moment that multiplayer felt real was when they were able to share a session while work was happening.

他写道，当某件事需要另一个意见时，我们俩都能打开同一个线程。

He writes, When something needed another opinion, we could both open the same thread.

当代理需要的信息只有我们其中某个人知道时，那个人可以直接添加进去。

When the agent needed information one of us had, that person could add it directly.

不需要把对话复制到Discord里，解释发生了什么，再把答案带回来。

There was no need to copy the conversation into Discord, explain what happened, and then carry the answer back.

我们就在同一个上下文里工作。

We were working inside the same context.

听起来只是界面上的一个小改进，但它改变了你跟代理协作的方式。

That sounds like a small interface improvement, but it changes the way you collaborate with an agent.

会话不再是一个开发者和一个模型之间的私密对话。

The session stops being a private conversation between one developer and a model.

它变成了一份共享的工作，其他可信的开发者可以查看、引导或者接管。

It becomes a shared piece of work that another trusted developer can inspect, steer, or take over.

为了让大家感受一下实际中的差异有多大，Colin分享了他之前一直在搭建一个新开发服务器，但需要把这个项目交给别人的经历。

To get a sense of how big the difference is in practice, Colin shared how he had been working to set up a fresh development server, but needed to hand that project over to someone else.

他写道，通常这种交接需要把我知道的所有东西整理成一份文档或者一条长消息。

Normally, he writes, that kind of handoff requires assembling everything I know into a document or a long message.

比如为什么做了某些决定，哪些方法已经试过不行了，项目当前的状态如何，哪些细节只在我脑子里，代理已经学到了什么。

Why certain decisions were made, which approaches had already failed, which state the project was in, which details existed only in my head, what the agent had already learned.

相反，他写道，另一位开发者用我们共享的 agent 开了一个 thread。

Instead, he writes, the other developer started a thread with our shared agent.

我打开了同一个 thread，直接补充了缺失的上下文。

I opened that same thread and added the missing context directly.

agent、另一位开发者和我，都在同一个连续的记录上工作。

The agent, the other developer, and I were all working from one continuous record.

然后他们就顺利开始了。

Then they were off and running.

没有复制粘贴式的交接，也没有尝试重建私人的 agent 对话。

There was no copy-and-paste handoff and no attempt to reconstruct a private agent conversation.

会话本身就成了交接文档。

The session itself became the handoff document.

显然，这种新的多人协作环境带来了很多挑战。

Now obviously this new multiplayer-style environment brings up a lot of challenges.

涉及所有权、权限和访问控制的问题。

There are questions of ownership and authority and access.

正如 Colin 所说，这还在早期阶段，我们也是这么对待的。

And as Colin puts it, this is still early and we're treating it that way.

不过，他写道，这个方向令人兴奋。

Still, he writes, the direction is exciting.

大多数开发者与 agent 的工作流仍然假设一个开发者、一个终端、一个私人对话。

Most developer-agent workflows still assume one developer, one terminal, and one private conversation.

最终的代码或许会被共享，但到达那一步的过程仍然藏在各个独立会话里。

The final code may eventually be shared, but the process of getting there remains hidden inside individual sessions.

多人协作的 agent 工作区让这个过程变得协作化。

A multiplayer agent workspace makes that process collaborative.

另一位开发者可以看到工作内容，理解上下文，补充自己所知道的，然后从之前停下的地方继续。

Another developer can see the work, understand the context, add what they know, and continue from exactly where it stopped.

没有转录导出，没有断掉的交接，也不用从零重建上下文。

No transcript dump, no broken handoff, no rebuilding the context from scratch.

只有一个共享的空间，开发者和 agent 可以继续推进工作。

Just one shared place where the developers and the agent can keep the work moving.

对我来说，有意思的是，这又是一次 OpenClaw 抢先走到了我们接下来要去的地方。

Now what's so interesting about this to me is that I think it is once again an example of OpenClaw getting to the place that we're going to head next before the rest of us.

就在我录这期节目的时候，我的 coding agents 正在后台开发下一个免费的 AI Daily Brief 学习体验。

Even as I record this, in the background, my coding agents are working on the next free AI Daily Brief learning experience.

那个不只是关于新的个人技能，而是关于一种新的方式来构建在团队层面运作的 agents。

And that one is not just about new individual skills, but a new way of building agents that operate at the team level.

如果你看你公司在做的所有工作，它会有两种形式。

If you take all the work you do inside your company, it's gonna come in 2 forms.

一个是你独自完成的工作，另一个是你和别人一起完成的工作。

Work you do alone, and work you do with others.

到目前为止，agents 实际上只是为独自完成工作而设计和启用的。

So far, agents have only really been designed and enabled for work you do alone.

然而，我们工作中很大一部分是需要一起完成的。

And yet, a huge portion of our work is work we do together.

我认为这就要改变了，我认为这是 agents 的下一个重大发展。而且我觉得，即使你不打算长期使用 OpenClaw，看看他们对 multiplayer 的思考方式也可能带来一些新想法。

I think that's about to change, I think that's the next big development for agents, and I think once again, even if you are not planning on being an OpenClaw user long-term, checking out the way that they're thinking about multiplayer might unlock some new ideas.

关于那个项目，很快就会详细介绍。但今天就到这里，这就是今天的 AI Daily Brief。

More on that project soon, but for now, that is gonna do it for today's AI Daily Brief.

一如既往，感谢你的收听或收看。

I appreciate you listening or watching as always.

下次再见，peace。

And until next time, peace.

Peace。

Peace.

已剔除 4 处广告（点击展开查看）

M10:44广告 · 已剔除

All right, friends, quick announcements before we dive in.

First of all, thank you to today's sponsors, KPMG, Blitzy, Section, and HyperAgent.

To get an ad-free version of the show, go to patreon.com/AIDailyBrief, or you can subscribe on Apple Podcasts. To learn more about sponsoring the show, send us a note at [email protected].

And to learn more about what we got cooking in the community, also check out aidailybrief.ai.

For example, you can find a link there to our next Agent Training for Executives program, which is coming up just after Labor Day.

Registration for that is open now.

M110:56广告 · 已剔除

If you're leading AI inside an enterprise, you already know that the gap right now isn't capability but execution.

That's why KPMG's You Can with AI is back with a new season featuring conversations with leaders like Sarojit Chatterjee of Emma.

Habib of Ryder, McKesson CIO Ellery Fisher, and others focused on practical execution.

What's working, what's not, and what it actually takes to move from pilots to real scaled impact across strategy, data readiness, governance, workforce, and value.

M111:25广告 · 已剔除

Go listen and subscribe at www.kpmg.us/aipodcasts.

That's www.kpmg.us/aipodcasts.

Blitzy deeply understands your codebase before it writes code.

Here's the first place that pays off: security in the age of AI.

Vulnerabilities don't live in isolation.

They live buried inside millions of lines of interconnected code, where patching one thing quietly breaks 3 others.

That's why surface-level scans fail.

Blitzy starts from its knowledge graph of your entire application, identifies and surfaces CVEs across the full estate, proactively recommends patches, and can execute the PR.

Each fix is grounded in how your systems connect and validate so nothing new breaks, and the knowledge graph dynamically updates, keeping you ahead of an ever-accelerating threat landscape.

One Blitzy customer resolved 21 active CVEs across 6 core microservices in 4 days.

Zero compile errors, every validation scan clean, months of planned work fixed in less than a week.

Security remediation grounded in real architectural context at the speed of compute.

Harden your codebase at blitzy.com.

That's blitzy.com.

Here's a harsh truth: your company is probably spending thousands or millions of dollars on AI tools that are being massively underutilized.

Half of companies have AI tools, but only 12% use them for business value.

Most employees are still using AI to summarize meeting notes.

If you're the one responsible for AI adoption at your company, you need Section.

Section is a platform that helps you manage AI transformation across your entire organization.

It coaches employees on real use cases, tracks who's using AI for business impact, and shows you exactly where AI is and isn't creating value.

The result?

You go from rolling out tools to driving measurable AI value.

Your employees move from meeting summaries to solving actual business problems.

And you can prove the ROI.

Stop guessing if your AI investment is working.

Check out Section at sectionai.com.

That's S-E-C-T-I-O-N-A-I dot com.

This episode of the AI Daily Brief is brought to you by HyperAgent, where you run fleets of agents your team can manage together.

M113:27广告 · 已剔除

HyperAgent deploys always-on agents in the cloud, doing real work across the tools your team already uses.

Marketing agents turn competitor moves into landing pages.

Sales agents enrich leads, draft emails, and update the CRM.

Ops agent chases the paperwork and tracks the budget.

Every agent has access to shared context and follows your rules about scope and approvals.

It's time you had agents that feel like teammates.

Hire yours at HyperAgent.

Get $100 in credits at hyperagent.com/AIDailyBrief.