科技热门推文 - 2026年8月18日

今日科技领域，人工智能基础设施和开发者工具成为焦点。Cursor 推出集成式代码托管服务，OpenAI 则寻求大规模扩充算力，并提升代码生成的安全性；Grok 也进一步增强了对开发者的吸引力。据报道，Anthropic 的业务增长以及通义千问开源模型的发展势头加剧了市场竞争压力。与此同时，初创企业吸引了大量资本，其中 Higgsfield 完成了4亿美元融资，规模居首。随着剪映 CapCut 的推动，AI 视频技术变得更加普及，但企业级应用成熟度、专有训练数据，以及依赖模型的“套壳”业务能否持续发展，仍是行业关注的核心问题。

1. Scobleizer (Group Score: 487.1 | Individual: 45.5)

Cluster: 22 tweets | Engagement: 2748 (Avg: 144) | Type: Tech

RT @cursor_ai: Origin, our code hosting platform, is now live.

It's fast, easy to use, and deeply integrated with Cursor.

Get started by syncing your repos from GitHub. https://t.co/aqRHavAOQg

See 21 related tweets

@XFreeze: Cursor just launched one of its biggest updates yet: Origin

Cursor can now HOST your code itself

T...

@ns123abc: Cursor just killed GitHub LMAO https://t.co/o33ODOdfqP\n\nQT @cursor_ai: Origin, our code hosting pl...
@cryptopunk7213: cursor went from being a coding wrapper to a full stack frontier AI lab building models, pioneering ...
@martin_casado: Finally!! And perfectly timed launch 🥳\n\nQT @cursor_ai: Origin, our code hosting platform, is now l...
@manthanguptaa: They were going to release it 2hrs back but github was down\n\nQT @cursor_ai: Origin, our code hosti...

2. yacineMTB (Group Score: 381.4 | Individual: 57.7)

Cluster: 10 tweets | Engagement: 1736 (Avg: 201) | Type: Tech

RT @DavidSacks: Some thoughts on Dario’s post:

Dario does not actually address Gavin Baker’s account of what he said – something he could easily deny if it were inaccurate.
Dario claims his critics live in a “bubble” where all regulation equals regulatory capture. He calls this an overly simplified view and notes that “Many people outside this bubble think of regulation as something that constrains corporate power and benefits ordinary people.” This argument is a straw man. Of course treating all regulation as capture would be overly simplified – but almost no one holds that view. I have repeatedly argued for strong antitrust enforcement to keep industries competitive, especially Big Tech. If Anthropic continues toward monopoly or duopoly status, I would be among the first to demand those rules apply.
Regulatory capture is not vague or in the eye of the beholder. Nobel laureate George Stigler defined it as regulation acquired by an industry and designed and operated primarily for its benefit. Stigler challenged the traditional view that government regulation arises from a benevolent state protecting the public from market failures. Rather, industry groups have concentrated stakes and pour resources into influencing regulators, whereas the public’s stake is diffuse and unorganized. The revolving door between companies and the agencies that regulate them compounds the problem. Anthropic understands these dynamics: it has hired multiple senior Biden AI-policy officials and built a substantial government-affairs operation plus a network of aligned organizations to push its preferred frameworks at state and federal levels.
Dario has consistently pushed for a new federal agency to review and approve frontier models prior to release – a proposal framed variously as an “FDA for AI,” an “FAA for AI,” and most recently a “FINRA for AI.” I call it a “DMV for AI” because a review process modeled on the FAA or FDA (which takes years) or FINRA (which issues rules for a staid industry widely seen as protecting incumbents) will create long queues as AI models wait for testing and approval. This process will only become more labyrinthine as rules accumulate to prevent theoretical harms. This would handicap the U.S. relative to China, which will not adopt the same constraints. It would also undermine Anthropic’s own business model, whose pricing power depends on remaining ahead of open models. Whatever Dario states today, it is difficult to believe the company would simply accept outcomes that erase that advantage.
Anthropic is on track to become one of the most valuable companies in history, with the resources to navigate any approval process and shape the rules while competitors wait. Dario wants open models under heavier scrutiny – he has called them dangerous in Senate testimony, criticized them for not being centrally monitored or withdrawn, and linked them to IP theft. He says he has never sought a ban, but he could achieve a similar result by insisting that identical rules apply to both open and closed models. The U.S. risks becoming an island of costly closed models while the rest of the world races ahead with broader choice.
Dario acknowledges that AI is structurally centralizing but attributes this mainly to chips and scaling laws. Access to compute matters, but the deeper risk is who decides which capabilities are available to whom. His preferred pre-deployment testing and FAA/FINRA-style oversight would place that gatekeeping power in a federal bureaucracy working hand-in-glove with a small number of frontier labs – reinforcing centralization rather than countering it.
The second part of Dario’s post assumes we have amnesia about Anthropic’s well-orchestrated campaigns hyping AI fears. His May 2025 claim that AI would wipe out 50 percent of entry-level knowledge jobs within five years still lacks supporting evidence fifteen months later. Similarly Anthropic breathlessly promoted its heavily contrived “blackmail” study on 60 Minutes. Yet Dario blames public negativity on a long-standing loss of trust in institutions rather than his own messaging.
These narratives have done more than anything to shape public fear. People are left asking the same question Mark Zuckerberg posed: why race to build a future you describe in such negative terms? Thomas Sowell’s "The Vision of the Anointed" captures the mindset – elite intellectuals convinced that only they are enlightened enough to control the outcome. As Zuckerberg notes, concentrating power in the hands of an enlightened few has rarely produced the promised results; the practitioners turn out to be less enlightened in practice than in self-conception.
Gavin Baker summarized the disagreement cleanly on our pod: Dario believes frontier AI is too powerful to distribute; we believe it is too powerful to centralize. Dario appears to believe, sincerely, that safety and progress are best served by centralizing authority in a marriage of corporate and state power. The weight of human history gives us reason to fear that outcome.

See 9 related tweets

@DavidSacks: Some thoughts on Dario’s post:

Dario does not actually address Gavin Baker’s account of what he ...

@WesRoth: People have heard:

AGI will transform science. AI will cure disease. Productivity will explode. Eve...

@rohanpaul_ai: A must read post, from @DavidSacks

"The U.S. risks becoming an island of costly closed models whil...

@BrianRoemmele: Well stated by @DavidSacks:\n\nQT @DavidSacks: Some thoughts on Dario’s post:

Dario does not act...

@TaylorLorenz: “Dario believes frontier AI is too powerful to distribute; we believe it is too powerful to centrali...

3. alex_prompter (Group Score: 281.0 | Individual: 37.7)

Cluster: 8 tweets | Engagement: 59 (Avg: 37) | Type: Tech

The agent you built for yourself isn't an agent. It's a hobby.

It runs on your laptop, carries your logins, and dies the second you close the lid. Your teammates can't run it. IT never approved it. It works as long as you personally babysit the loop.

@xpander_ai just launched Omni to end that. Publish an agent once, teammates who are cleared just ask it, no ticket. Runs in your cloud or air-gapped on your own GPUs, on GPT, Gemini, Llama, Qwen, or your own fine-tune, and you can swap models mid-task when a cheaper one will do.

Permissions scoped per user. Actions logged. Built to walk into a security review instead of dying in one.

Their published GAIA score: 90.9% on long multi-system tasks. Full methodology and results are open on GitHub.

Congrats @dudutwizer and team. Honest question for my builder friends: how many of your agents survive you taking a week off?\n\nQT @dudutwizer: We've raised $7.5M to democratize AI agents and help every business become AI-native, without the vendor lock-in.

Most companies face a binary choice: lock your entire AI strategy into one vendor, or fragment across dozens of point solutions that become operational nightmares.

The reality is that your teams are already using desktop agents, with or without your approval.

xpander gives you control and ownership on your AI stack , with full flexibility and governance without compromising performance.

Try the cloud version with 1000 free credits: https://t.co/gXmoRmzuIH today

Or schedule a demo for the self-hosted version.

See 7 related tweets

@aakashgupta: xpander just raised $7.5M for the least fashionable position in enterprise AI. Neutrality.

Every ma...

@kimmonismus: Enterprise AI figured out a second business model: sell the platform, then rent out the humans who m...
@Scobleizer: Watch this.

They literally made a Mac vs PC commercial… but for AI agents.

Left side: guy builds h...

@wallstengine: XPANDER AI LAUNCHES OMNI

A built-in AI agent that builds, runs and governs other agents across a co...

@StockSavvyShay: XPANDER AI built a system that creates runs and governs other agents by treating frontier models as ...

4. alexcooldev (Group Score: 231.4 | Individual: 32.0)

Cluster: 11 tweets | Engagement: 98 (Avg: 105) | Type: Tech

People are still arguing that AI wrappers are dead and any product that doesn’t build its own model will eventually get killed by Claude...

Meanwhile, OpenRouter, basically an API gateway for AI models, just got acquired by Stripe for $7B+. 🥴

You don’t always need to build the model.

Sometimes the biggest opportunity is building the infrastructure, distribution, or product layer around it. 😌

See 10 related tweets

@WesRoth: Stripe has reportedly finalized an agreement to acquire OpenRouter for more than $7 billion, a massi...
@business: Stripe has finalized an agreement to acquire OpenRouter, a startup that helps companies switch betwe...
@IPONewsroom_: STRIPE IS REPORTEDLY BUYING OPENROUTER FOR MORE THAN $7 BILLION, PER BLOOMBERG

That's roughly 5x Op...

@VaibhavSisinty: Stripe is reportedly acquiring OpenRouter for over $7 billion.

82 days ago, OpenRouter raised its S...

@cryptopunk7213: stripe acquiring openrouter makes a lot of sense once you realize they both do the same thing: token...

5. wallstengine (Group Score: 149.0 | Individual: 34.0)

Cluster: 6 tweets | Engagement: 182 (Avg: 167) | Type: Tech

ANTHROPIC REVENUE RUN RATE JUMPS 7X AHEAD OF IPO

July 2026: $65B+ May 2026:$ 47B End of 2025: $9B

Anthropic also reported more than $11.5B in preliminary revenue for its latest completed quarter, up from$ 787M a year earlier, while posting positive adjusted operating income.

The company has confidentially filed to go public and could make its Wall Street debut as soon as this fall.

See 5 related tweets

@kimmonismus: Anthropic’s annualized revenue reportedly hit $65 billion by the end of July, more than 7x its run r...
@FirstSquawk: ANTHROPIC'S REVENUE RUN RATE HAS SURPASSED $65 BLN AHEAD OF ITS IPO, WITH THE COMPANY ON COURSE TO G...
@negligible_cap: *ANTHROPIC REVENUE RUN RATE SURPASSES $65 BILLION AHEAD OF IPO\n\nQT @negligible_cap: *ANTHROPIC PRE...
@Techmeme: Sources: Anthropic's revenue run rate reached $65B by the end of July, up from$ 47B in May and $9B i...
@axios: Anthropic's revenue run rate reportedly surpasses $65 billion pre-IPO https://t.co/WHqgIgs77r...

6. dejavucoder (Group Score: 140.5 | Individual: 32.4)

Cluster: 5 tweets | Engagement: 65 (Avg: 40) | Type: Tech

google is rl env maxxing\n\nQT @abhijaymrana: Wow, seems like Google is buying Spirit Airlines' enterprise data for $10m (outbidding Mercor at$ 7.5m).

Basically includes every internal document, email, workflow, and codebase for a once $6B company.

Honestly, $10m for 34 years of operational data really seems like a steal. https://t.co/9qin2ef6Az

See 4 related tweets

@hammer_mt: Your company data just isn't worth that much –– $20 per worker per year.\n\nQT @abhijaymrana: Wow, s...
@every: Tea leaves:

The value of data is shockingly low.

All of it—internal Slack, emails, every message e...

@rickasaurus: RT @abhijaymrana: Wow, seems like Google is buying Spirit Airlines' enterprise data for $10m (outbid...
@BusinessInsider: In a competitive AI data race, Google buys Spirit Airlines' corporate data for $10 million, eclipsin...

7. testingcatalog (Group Score: 133.9 | Individual: 40.3)

Cluster: 4 tweets | Engagement: 361 (Avg: 259) | Type: Tech

OpenAI is planning to secure 8 gigawatts of IT at the PORTS-Pike Technology Campus, working with SB Energy, NVIDIA, and the U.S. Department of Energy.

An initial combined community investment is $80 million, with$ 40 million from OpenAI.

OpenAI will also make up to $84 million in Codex credits available to approximately 844,000 eligible Ohio students.\n\nQT @OpenAINewsroom: OpenAI has entered into an agreement to utilize capacity at the PORTS-Pike Technology Data Center in Pike County, Ohio, working with SB Energy, NVIDIA, and the U.S. Department of Energy.

Pike County helped power American industry in the last century. Here’s how we’ll ensure this project becomes a catalyst for future jobs, investment, and long-term opportunity in Southern Ohio.

SB Energy will pay the full cost of all grid upgrades and new transmission lines needed to serve the data center so the costs aren’t shifted to Ohio or other ratepayers in the region.
This data center will use closed-loop, air-cooled cooling systems that recirculate water rather than relying on cooling towers that continuously consume water. It is expected to use significantly less water than was historically used by the Portsmouth gaseous diffusion plant.
The project is expected to create 35,000 construction jobs during its six-year buildout through 2032 and 2,500 long-term operating jobs, with a focus on local workers, contractors, suppliers, trades, and service providers.
We will enable an initial combined community investment of $80 million:$ 40 million from OpenAI for priorities defined by the local community alongside SB Energy’s existing $40 million commitment. The project is also expected to generate hundreds of millions of dollars in state and local tax revenue over its lifetime, with local property taxes primarily supporting public schools.
OpenAI will make up to $84 million in Codex credits available to approximately 844,000 eligible Ohio college, community college, and technical school students aged 18 and older during the 2026–2027 academic year.
Each year, we will publish a report on our progress, and local input will help shape the Ohio Community Compact.

https://t.co/uhVTGVCYpz

See 3 related tweets

@kimmonismus: OpenAI has entered an agreement to secure approximately 8 GW (!) of AI capacity in Ohio, one of its ...
@WesRoth: OpenAI has formally committed to roughly 8 gigawatts of AI compute capacity at the PORTS-Pike Techno...
@wallstengine: $NVDA BACKS OPENAI’S OHIO AI CAMPUS

Nvidia will invest $1.5B in SB Energy and support an initial 4....

8. omarsar0 (Group Score: 126.7 | Individual: 35.9)

Cluster: 4 tweets | Engagement: 52 (Avg: 80) | Type: Tech

Grok Bot is so good! I am very impressed with all the recent updates on Grok models as well.

If you ever wanted to try a different product outside Codex and Claude Code, Grok Bot is worth trying.\n\nQT @GavinSBaker: I think @bot is another “Claude Code” moment for AI.

I would estimate my personal AI usage is up something like 100x.

And for everyone who reached out about how to build a “podcast summarizer” it took me about 15 seconds in Grok Bot and is better than what I had before. https://t.co/AaoeHbEqtl

See 3 related tweets

@elonmusk: Grok @Bot\n\nQT @GavinSBaker: I think @bot is another “Claude Code” moment for AI.

I would estimate...

@ns123abc: i’ve been saying for a week that @bot is the best new app since chatgpt dropped in 2022

people lau...

@levelsio: RT @GavinSBaker: I think @bot is another “Claude Code” moment for AI.

I would estimate my personal ...

9. EmmaUsesAi (Group Score: 122.6 | Individual: 29.8)

Cluster: 7 tweets | Engagement: 40 (Avg: 56) | Type: Tech

CapCut App is making it easier to try Seedance 2.5.

Get official access with zero wait, starting from $0.06/sec.

Simple setup, powerful AI video generation, all in one app.

https://t.co/M47CVRwUl6

#CapCut #CapCutApp #Seedance25 #CapCutai #CapCutDidThat https://t.co/2hKVtFCfoo

See 6 related tweets

@NovaIAHQ: RT @OliviaReedai: The ultimate AI upgrade has arrived. 🚀

Experience the power of Seedance 2.5 with...

@NeuraFlowAix: 1 App. Top AI Model. Lowest Price. 🥇

There is literally no better way to use Seedance 2.5 than on ...

@NovaIAHQ: RT @Lacoste0x_: Stop paying premium prices for slow AI tools. 🐢

CapCut offers Seedance 2.5 at a fr...

@NovaIAHQ: RT @TheoBuildsAI: CapCut App is a simple place to start experimenting with Seedance 2.5.

You get of...

@NovaIAHQ: RT @TheAva_AI: CapCut App gives you direct access to Seedance 2.5.

Official Seedance 2.5 access · Z...

10. rohanpaul_ai (Group Score: 114.8 | Individual: 32.8)

Cluster: 6 tweets | Engagement: 39 (Avg: 41) | Type: Tech

Greg Brockman says OpenAI is developing models that can write “superhumanly secure code” to make institutional infrastructure tougher against AI-powered attacks.

"we are starting to train our models specifically to write superhumanly secure code.”

“Over the coming months, every organization will need to begin significantly automating its security program to stay secure, and the security community must urgently rise to define the tools, practices, and playbooks that will increase the power of defenders faster than that of attackers as AI continues to advance.”

“Keep human review for consequential changes, but eliminate the unnecessary delay between identifying a real problem and putting a safe fix in front of an engineer.”\n\nQT @gdb: defenders can see the future, and have a narrow window to uplevel their cybersecurity practices now.

key is to uplevel fundamentals and apply the best AI tools.

what we’re doing at OpenAI, and where other organizations can start: https://t.co/P3IMZkV234

See 5 related tweets

@laurenmhreeder: It's clear AI is currently scaling offensive capability faster than its defensive capacity.

@dan_l...

@danshipper: impt read:\n\nQT @gdb: defenders can see the future, and have a narrow window to uplevel their cyber...
@dkundel: RT @gdb: defenders can see the future, and have a narrow window to uplevel their cybersecurity pract...
@Techmeme: Greg Brockman calls the OpenAI-Hugging Face incident "a watershed moment" and discusses how OpenAI a...
@zerohedge: *OPENAI: CONTINUE TO INVEST IN SECURE ARCHITECTURE & CONTROLS

*OPENAI: PUTTING MODELS TO WORK D...

11. mollytaft (Group Score: 93.7 | Individual: 47.9)

Cluster: 4 tweets | Engagement: 11876 (Avg: 735) | Type: Tech

RT @jason_koebler: Exclusive: A rare bookseller suspected his books were being bought by AI companies. He stuck an AirTag in a book and we tracked it to a secretive Amazon facility in Las Vegas where they scan and destroy books called VGT3:

https://t.co/WXSo0zB6JF

See 3 related tweets

@Mayhem4Markets: Amazon started as a bookseller. Now it's buying rare books by the thousand, slicing off their spines...
@Techmeme: Investigation: Amazon is buying huge quantities of rare books, scanning them for AI, and destroying ...
@arstechnica: Hidden Airtag reveals Amazon is trashing rare books to train AI https://t.co/B0b1WHBtfU...

12. aakashgupta (Group Score: 93.3 | Individual: 35.5)

Cluster: 3 tweets | Engagement: 53 (Avg: 54) | Type: Tech

Alibaba's Qwen just out-downloaded Meta's open models 9 to 1 this year on Hugging Face, the platform Llama built its reputation on. Google did 418 million downloads in 2026. Meta did 227 million. Qwen did 2 billion, and the download count is the least interesting number in the report.

The interesting one is 300,000. That's how many derivative models developers have built on top of Qwen's 460 open releases. Every fine-tune is a team that trained on Qwen's architecture, learned Qwen's quirks, and shipped something that breaks if they switch. Downloads churn. Derivatives lock in.

Meta wrote this playbook. Llama is the reason "open weights" became a strategy instead of a hobby. Then the company turned toward closed frontier work right as the flywheel started compounding, and Alibaba picked the position up off the floor.

There's cold logic behind the giveaway. Qwen ships through Alibaba Cloud into Southeast Asia and Africa, markets where a free, customizable model beats a metered API on price before the sales call even starts. The model costs Alibaba money. The cloud contracts underneath it are the business.

Washington restricted the chips. Alibaba answered by giving away the software, and the software layer is where developers actually standardize.

Meta will spend up to $145 billion on AI infrastructure this year. Its open models got 227 million downloads. The company it's chasing got 2 billion, plus 300,000 teams building on top.\n\nQT @Polymarket: JUST IN: Alibaba’s Qwen becomes the world’s No. 1 open AI model by downloads, topping 3 billion globally.

See 2 related tweets

@WesRoth: The most important Qwen number probably isn't 3 billion downloads.

It's 151,448 derivative models o...

@Ric_RTP: RT @Ric_RTP: The most downloaded AI on earth is now Chinese.

Alibaba just gave away a model that ma...

13. wallstengine (Group Score: 93.3 | Individual: 24.4)

Cluster: 6 tweets | Engagement: 193 (Avg: 167) | Type: Tech

HIGGSFIELD RAISES $400M AT$ 5.4B VALUATION

AI video startup Higgsfield raised a $400M Series B from investors including Goldman Sachs, Intel and DST Global, valuing the 2-year-old company at$ 5.4B.

Annualized revenue hit $700M in August, up from just$ 20M a year ago, with 30M+ users across 238 countries and territories. Most revenue now comes from businesses using its AI tools for marketing content.

See 5 related tweets

@Techmeme: AI video generation startup Higgsfield raised $400M from DST, Goldman Sachs, Liberty Global, Intel, ...
@MattSwulinski: $400M at$ 5.4 B.

Strip the AI label and Higgsfield is a DTC brand.

Timed giveaways, daily drops,...

@SquawkCNBC: AI video creation platform @higgsfield has reached a valuation of $5.4B after its latest funding rou...
@alex: RT @etnshow: Higgsfield just raised $400M at a$ 5.4B valuation.

CEO @alexmashrabov on what this act...

@Scobleizer: RT @fttechnews: Higgsfield valued at $5.4bn as Goldman and Intel back AI video start-up https://t.co...

14. ANI (Group Score: 90.5 | Individual: 31.9)

Cluster: 3 tweets | Engagement: 5 (Avg: 141) | Type: Tech

#WATCH | Delhi: Union Minister Ashwini Vaishnaw says, "India has achieved complete self-reliance in several types of components. For instance, the entire domestic demand for enclosures is now met within the country—India is 100% self-reliant in this area. We are also meeting 45% of the demand for PCBs (Printed Circuit Boards) and have even begun exporting them. India is currently meeting 60% of the domestic demand for lithium-ion cells—used in digital applications. We have achieved a level of 80% self-reliance in laminates and have started exporting anode materials. Regarding optical transceivers, we have moved beyond merely meeting 350% of the demand to achieving complete self-reliance and commencing exports. Similarly, 75% of the demand for connectors is being met domestically. We have become fully self-reliant in relays and have started exporting them. The resolve to modernize, expand, deepen, and broaden India's electronics industry—and to manufacture components, materials, and machinery across the entire spectrum—will lay a strong foundation for the nation... There were critics who constantly created obstacles and voiced skepticism, claiming India could not become a manufacturing nation. Today, a clear blueprint of success stands before them... Regarding the scale of production, these 106 projects under the scheme are projected to generate a total production value of Rs 5,34,000 crore. It will bring GST revenue to the country, generate income tax, create employment, develop supply chains, and establish new quality benchmarks—all within the country..."

See 2 related tweets

@ANI: #WATCH | Delhi: Union Minister Ashwini Vaishnaw says, "Electronics manufacturing has grown sevenfold...
@ANI: #WATCH | Delhi: Union Minister Ashwini Vaishnaw and MoS Jitin Prasada attend an event for the handov...

15. TipRanks (Group Score: 89.7 | Individual: 34.4)

Cluster: 3 tweets | Engagement: 39 (Avg: 48) | Type: Tech

🔼Apple $AAPL upgraded to Buy from Neutral at Rothschild & Co Redburn

Rothschild & Co Redburn analyst Timm Schulze-Melander upgraded Apple to Buy from Neutral with a price target of $400, up from$ 260.

The firm sees iPhone sales growing at 12% annually, or up to 14% above consensus expectations, through fiscal 2030, led by Apple's entry into the premium foldable market with the launch of the iPhone Ultra. Stronger than consensus iPhone earnings and a "strategic reset" to Apple Intelligence leave Rothschild's estimates 18% above fiscal 2030 consensus.

It has confidence that Apple's relative valuation multiple is "well supported." The firm believes Apple may be able to "assert itself as the gatekeeper of consumer AI, delegating AI workloads to a subservient fleet of open-source models, monetized via Apple Pay and other payment platforms."

See 2 related tweets

@wallstengine: Rothschild Redburn Upgrades $AAPL to Buy, PT$ 400

Analyst comments: "Apple’s defining characteristi...

@DeItaone: $AAPL - APPLE UPGRADED TO BUY —$ 400 TARGET

Rothschild Redburn upgraded Apple from Neutral to Buy...

16. WesRoth (Group Score: 86.6 | Individual: 31.6)

Cluster: 3 tweets | Engagement: 64 (Avg: 36) | Type: Tech

Anthropic is “ramping up its efforts very quickly in biology and medicine,” hopes for major results over the coming years, and expects some “early glimmers” in the coming months.

• Claude Science gives researchers an AI workbench for scientific computing and research, with an early focus on biology and biomedicine. Anthropic is supporting up to 50 projects with as much as $30,000 in Claude credits each.

• Its broader AI for Science program provides Claude access to researchers working on areas including biological systems, genetics, drug discovery, and other high-impact science.

• Anthropic recently launched a rare-disease initiative offering selected researchers and early-stage biotechs up to $50,000 in Claude credits over six months.

• The company has also been developing evaluations such as BioMysteryBench to measure whether Claude can analyze messy real-world biological datasets rather than simply answer textbook questions.\n\nQT @DarioAmodei: 2/2 Second, on the messaging around AI. I do not agree that my messaging has been disproportionately negative. In fact it has been about equally balanced between risks and benefits: I’ve written one major essay about each, and even in interviews where I discuss the risks, I make sure to frequently mention the incredible benefits as well as proposing possible solutions to the risks (short clips from my interviews that end up on social media tend to be disproportionately negative, as that gets clicks). In fact, I wrote Machines of Loving Grace because I didn’t feel the AI industry was painting an inspiring enough picture of how the technology could radically transform the world for the better. The bulk of the essay is devoted to refuting skepticism of AI’s potential in health and biology, and showing why I think it will actually be possible to cure most human disease in ~5-10 years, as crazy as it may sound to ordinary people and frankly to biologists as well (I used to be one!). And, if you read my most recent essay (Policy on the AI Exponential), I discuss concrete proposals for how to streamline the FDA process to make sure the deluge of AI-accelerated drugs isn’t slowed down by the regulatory process. I feel the urgency here: I lost my father to Hepatitis C only a few years before the development of direct-acting antivirals (sofosbuvir), which cure 95% of patients and probably would have cured him.

I do agree that the public has a negative view of AI (and that this is a big problem), but I don’t think it is primarily caused by me or any other AI leader warning about AI’s risks. I think it is fundamentally a crisis of trust. I think that ordinary people don’t trust companies, governments, or the tech industry and always suspect that we are cooking up some new way to screw them over. The causes of this go back decades and AI is just the latest iteration of it. I don’t think that a glitzy marketing campaign with a positive spin (which some have advocated that Anthropic do) is the way to win back that trust — at this point, saying that AI will cure cancer is more a cliche than it is inspiring, and most people think it is deceptive. The thing that will work is actually curing cancer. I think by far the most accurate criticism of AI companies including Anthropic is that we haven’t yet delivered on our big promises to benefit the world. That is totally on us, and I think it’s the criticism you should be making, instead of all this stuff about messaging and marketing.

We are however doing our best to fix this: Anthropic is ramping up its efforts very quickly in biology and medicine, and we hope to have incredible results in the coming years and some early glimmers in the coming months. When we’ve actually accomplished something real, the whole world will hear about it, as loudly as possible, you have my word on that. But until then I don’t want to make empty promises, and in the meantime I feel compelled to speak honestly about the very real risks of AI and how to address them. Honesty is the right thing on the merits, and in terms of public credibility and trust it is no worse than, and may in fact be better than, an approach that ignores or distracts from risks which people instinctively understand are real.

See 2 related tweets

@VaibhavSisinty: Dario Amodei rarely posts on social media. Two days ago he did, and it's worth reading.

He replied ...

@hammer_mt: Nobody has cancer but nobody has a job either\n\nQT @DarioAmodei: 2/2 Second, on the messaging aroun...

17. indie_maker_fox (Group Score: 84.8 | Individual: 24.5)

Cluster: 6 tweets | Engagement: 100 (Avg: 19) | Type: Tech

朋友们，Mintlify 的开源平替来啦！

最近我发现一个很有意思的开源文档框架：Blume。

https://t.co/jiOGLqxJny

它和我之前开源的 MkDocs 一样，都是 Markdown-first，核心目标都是：少花时间搭文档站，多花时间写内容。区别在于 Blume 更激进地把导航、搜索、SEO、OG、i18n、OpenAPI Playground、AI/MCP 和 llms.txt 都做成了开箱即用。

如果说 MkDocs 更像可定制、可延展的文档站模板，Blume 则更像“只给我一文件夹 Markdown，其余交给框架”的生产级 Docs as Code 方案。很适合关注 AI 可读性与开发者体验的项目。\n\nQT @indie_maker_fox: 🎉 开源个小项目：MkDocs

简易的文档站模板，基于 FumaPress 框架，可以一键部署到 Cloudflare Workers，支持多语言、暗黑模式、YouTube视频组件、复制Markdown，以及自动生成sitemap、OG图片等功能。如果你要搭建文档站点，只需要编写mdx文档，配置网站名称、链接、域名即可。

https://t.co/7EpohXo3m2

MkDirs 和 TanStarter 模板文档站已使用 MkDocs 重新部署。

See 5 related tweets

@xicilion: 忍不了了，这条尬吹怎么会出现在我的时间线里的？\n\nQT @AYi_AInotes: 你永远可以相信@deepseek_ai 开源的诚意！！！ DeepSeek 竟然把自己内部工程团队吃饭的家伙直接...
@indie_maker_fox: RT @indie_maker_fox: 朋友们，Mintlify 的开源平替来啦！

最近我发现一个很有意思的开源文档框架：Blume。

https://t.co/jiOGLqxJny

它和我之...

@indie_maker_fox: RT @indie_maker_fox: 最近研究了下 Matt 的技能集合，也体验了用这套技能进行开发的流程。主流程如下：

1、grill-me / grill-with-docs：开工前先拷问你...

@indie_maker_fox: RT @indie_maker_fox: 我看很多人都在转发这个视频，蛮多人也提到 pi 的开发者的一个观点，pi 不需要记忆系统，为什么？因为他觉得，code 是 source of truth，增...
@indie_maker_fox: RT @indie_maker_fox: 推荐一个制作 PPT 的好工具 Bento. page

我试了好几个 AI 制作 PPT 的工具，好几个都翻车了，目前感觉这个体验最好，基于之前分享的 Pi...

18. IBuzovskyi (Group Score: 82.8 | Individual: 29.3)

Cluster: 3 tweets | Engagement: 58 (Avg: 92) | Type: Tech

HERMES AGENT SHIPPED A WAVE OF UPDATES THIS WEEK. /LOOP COMMAND. FASTER DESKTOP. BIGGER CONTEXT WINDOWS. HERE IS EVERYTHING THAT CHANGED.

hermes update

/LOOP (the headline feature)

new command. re-runs your prompt on repeat inside the same session.

/loop 2m check the deploy status

every 2 minutes Hermes wakes up. reads the state fresh. reports back. repeats until the task is done or you type /loop stop.

set the interval yourself: /loop 30s watch the error log /loop 5m monitor the queue --until depth zero /loop 1m poll CI --times 30

or let Hermes decide (self-paced): /loop keep an eye on the migration

starts checking every minute. backs off when nothing changes. snaps back when something does.

full /loop deep-dive in the article below.

DESKTOP SESSIONS LOAD FASTER

sessions in Hermes Desktop were slow to open. long conversations took seconds to render. fixed. sessions load significantly faster now.

if you had a 200-turn session that froze when you switched back to it, that's gone.

CODEX USERS: CONTEXT WINDOW UP TO 350K

if you use Hermes with a Codex subscription, context window is now 350K tokens.

if OpenAI exposes the full 1M for Codex, Nous Research will open it up immediately.

CHATGPT SUBSCRIPTION USERS: 900K CONTEXT

if you run Hermes through your ChatGPT OAuth, you now have 900K tokens of context. enough to hold entire codebases in a single session without compression.

STABILITY PATCH (v0.20.2, August 16)

Desktop: multi-gateway Connections registry, profile-scoped refreshes, MCP health checks. CLI: Windows update probes, Kitty keyboard protocol. Gateway: persisted model routes, /loop completion, Telegram DM topics. Caching: prompt caching for LiteLLM Claude on the OpenAI wire. Cron hardened. Auth hardened. Installer hardened.

hermes update\n\nQT @IBuzovskyi: https://t.co/CPnDM0ndot

See 2 related tweets

@iamlukethedev: Hermes merged 132 PRs today

It seems that the focus is more on the Desktop app this month.

Hopeful...

@TeksEdge: Have you updated your Hermes Agents with this amazing cleanup release yet? Why not?\n\nQT @TeksEdge:...

19. daleverett (Group Score: 81.6 | Individual: 29.8)

Cluster: 3 tweets | Engagement: 4 (Avg: 9) | Type: Tech

Raising pre-seed? These 16 funds write first checks of $250K–$ 500K, most with no warm intro 👇

@PrecursorVC @HustleFundVC @fdotinc @Boldstartvc @pearvc @AforeVC @BoostVC @2048vc @GoAheadVC @forumventures @ldvcapital @redbudvc @outlandervc @ChargeVC @RightSideCapVC @actionscapital

Pre-seed investors aren't buying your revenue. They're buying your speed.

Founders who ship an agent in a weekend raise faster than founders still wiring up memory, retrieval, and 3 databases in month two.

That's literally why we built @polygres an OSS ai-native search layer for postgres (free for this week).

DM me for developer credits, or pitch me and I'll fund your infra.

Which fund did I miss? I'll add the best replies to the next list.

See 2 related tweets

@daleverett: If you're building an agent on @polygres pitch me and i'll fund your infra 👀\n\nQT @damienhe: Most f...
@damienhe: if you're building agentic systems for enterprise. pitch me. i'll fund ur infra\n\nQT @damienhe: Mos...

20. BytePlusGlobal (Group Score: 81.1 | Individual: 30.8)

Cluster: 3 tweets | Engagement: 38 (Avg: 100) | Type: Tech

Short-form video is already a billion-dollar business. The next wave of growth may come from AI-native production.

Catch the highlights from AZ8 Co-Founder × 2xLabs Creative Director: The Future of AI-Native Video Production, and see how platforms and creators are building more controllable, repeatable, and production-ready AI video workflows.

• Platform workflow insights from @AZ8studio • Creative practices from @2x_labs
• Dreamina Seedance 2.5 highlights presented by BytePlus — native 30-second clips, up to 50 multimodal reference assets, and targeted editing without starting over

Watch now.

See 2 related tweets

@heyshrutimishra: The film industry's real moat was never a studio lot.

It was the cost of consistency, keeping the ...

@Scobleizer: RT @AZ8studio: AI video is going through a bigger shift than model improvements alone.

We’re moving...