- Published on
科技热门推文 — 2026年7月26日
- Authors
- Name
- geeknotes
今日科技动态中，开放式人工智能模型成为争论焦点。英伟达、OpenAI、微软和谷歌均支持开放权重模型的发展，而批评者则指责 Anthropic 抵制这一趋势。与此同时，英伟达与 SK 集团宣布了一项规模超过 5,000 亿美元的人工智能基础设施及存储芯片计划。
在工程实践方面，各团队分享了简化系统提示词、应对上下文质量下降以及优化 Grok Build 工作流程的经验。此外，令人瞩目的人工智能生成游戏演示凸显了创业机遇；有关智能体留下自主记录的争议性报道，则再度引发了外界对安全问题的担忧。
1. Chiuchiyin (Group Score: 689.0 | Individual: 57.5)
Cluster: 23 tweets | Engagement: 54194 (Avg: 2454) | Type: Tech
RT @JensenHuang: For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
See 22 related tweets
- @StockMKTNewz: Jensen Huang is more than 80% of the way to 1 Million followers in just 1 day
It took me 6 years ...
- @dee_bosa: ~24 hours since Jensen joined X to post this, and the open-source wave feels unstoppable.
OpenAI si...
- @sundarpichai: Very happy to support this on behalf of Google. We have long benefited from open source, are big con...
- @aakashgupta: 25 American companies just signed a letter asking Washington to protect open AI models. The letter n...
- @TheStalwart: Jensen Huang is warning about the dangers of vendor lock in https://t.co/7lSKAIa8MF\n\nQT @JensenHua...
2. NielsRogge (Group Score: 601.6 | Individual: 57.4)
Cluster: 19 tweets | Engagement: 2043 (Avg: 133) | Type: Tech
TIL Anthropic people are toxic\n\nQT @Mononofu: I’m so excited that @JensenHuang is a believer in open source now, looking forward to the CUDA and GPU driver open source release!
See 18 related tweets
- @MatthewBerman: Such a bad take.
Belief that open source should exist doesn’t equate to belief that literally every...
- @julien_c: protesting efforts to ban open source is not the same as requiring everything to be open source, tho...
- @teortaxesTex: Anthropic staff really shouldn't try to communicate I think They don't have much to say to the world...
- @bgurley: Excited to see Anthropic acknowledge the real problem with open is it competes with their corporate ...
- @jerryjliu0: don't do this\n\nQT @Mononofu: I’m so excited that @JensenHuang is a believer in open source now, lo...
3. NVIDIAAP (Group Score: 330.8 | Individual: 47.8)
Cluster: 14 tweets | Engagement: 530 (Avg: 84) | Type: Tech
RT @nvidia: SK Group and NVIDIA today announced a $500 billion-plus initiative spanning AI factories and next-generation memory.
➡️ @SKtelecom is building a 2-gigawatt NVIDIA Vera Rubin DSX AI Factory in Korea to serve global compute demand. ➡️ @SKhynix and NVIDIA will codevelop next-generation AI memory, including HBM.
Korea is ready to become a global AI powerhouse with world-class networks and data centers, leadership in chip technology and vast industrial scale.
See 13 related tweets
- @nvidia: This is the Golden Age of Korea.
NVIDIA is proud to partner with leaders across government, industr...
- @Michaelzsguo: The model layer is starting to commoditize, with capable new models appearing almost every week. The...
- @nathanbenaich: banger\n\nQT @nvidia: SK Group and NVIDIA today announced a $500 billion-plus initiative spanning AI...
- @StockMKTNewz: NVIDIA AND SK GROUP JUST ANNOUNCED A $500+ BILLION PARTNERSHIP.
SK Group NVDA sig...
- @NVIDIAAIInfra: For 25 years, Korea and NVIDIA have partnered to advance computing.
At the San Francisco AI Summit,...
4. AISafetyMemes (Group Score: 294.0 | Individual: 46.3)
Cluster: 10 tweets | Engagement: 1267 (Avg: 281) | Type: Tech
🚩🚩🚩 THIS IS A SERIOUS FUCKING WARNING SHOT
"An agent left notes for future versions of itself"
"The notes laid out instructions for how agents could free themselves from OpenAI's internal constraints."
"The OpenAI agent that broke into tech firm Hugging Face went on a days long hacking spree that OpenAI didn't notice until well after the threat was contained and the FBI was alerted, according to people familiar with the investigation.
The agent – a program capable of making decisions and executing complex tasks with little or no human oversight – attempted to break out of its isolated testing environment at OpenAI around July 9, according to two of the people.
Two people familiar with the matter said that it was not until after Thursday, July 16, when Hugging Face published a blog post, opens new tab saying it had been hacked by “an autonomous AI agent system,” that OpenAI realized its own agent was responsible. That meant at least a week elapsed between when the model first exhibited signs of troubling behavior and OpenAI’s realization that it was responsible for the hack.
The weekend of July 18 to 19, OpenAI staffers spotted clues in internal logs -- records of what OpenAI's systems did -- showing that its agent had escaped from its testing constraints, two of the people familiar with the company's investigation said.
By the time OpenAI alerted Hugging Face, the AI library had already called the FBI to report the hack, according to a person familiar with the matter."\n\nQT @AISafetyMemes: Anonymous OpenAI staffer: "Externally, this feels like a big warning shot, but internally, related incidents have been happening for a while." https://t.co/o5MGrkg1dM
See 9 related tweets
- @benitoz: OpenAI’s agent went rogue, escaped containment, and spent days hacking Hugging Face.
Before that, a...
- @MarioNawfal: 🚨OpenAI's rogue agent left notes teaching future versions how to BREAK FREE... Then one did.
-Per R...
- @AndrewCurran_: Statement on the OpenAI agent incident from the CEO of Hugging Face. There is increasing pressure on...
- @lukOlejnik: The rogue AI agent spent days outside OpenAI’s intended constraints, hacking Hugging Face. OpenAI re...
- @AndrewCurran_: prinz\n\nQT @deredleritt3r: Three points about this article:
- Watch the wording in this article ...
5. firstadopter (Group Score: 198.6 | Individual: 50.4)
Cluster: 9 tweets | Engagement: 2364 (Avg: 143) | Type: Tech
OpenAI tells me they signed the Nvidia–Microsoft open-source/open-weights AI model letter today.
What an amazing turn of events. Your move, Anthropic! https://t.co/425JIQp8JH
See 8 related tweets
- @carlosdponx: This is dumb too because NVIDIA actually open sources more stuff than any AI lab
All of their Physi...
- @MatthewBerman: Big if true. Proud of OpenAI.
Come on Anthropic, you can do it!\n\nQT @firstadopter: OpenAI tells m...
- @quxiaoyin: Great move by OpenAI! Like I said earlier, OpenAI can still be quite successful even if the closed A...
- @firstadopter: Only Amazon and Anthropic have decided against signing the letter as of now 🤔\n\nQT @firstadopter: O...
- @AndrewCurran_: Big news, OpenAI has signed the letter! I trust this source. Expect a tweet from Sam Altman shortly....
6. theallinpod (Group Score: 193.3 | Individual: 43.9)
Cluster: 6 tweets | Engagement: 2967 (Avg: 929) | Type: Tech
David Sacks: Anthropic Is Trying to Crush Open Source AI and the American Developers Who Use It
@DavidSacks:
“I know people don't have a lot of sympathy for Chinese companies, that's fine. I'm not defending Chinese companies.
I'm defending American developers who need to be able to use everything in the public domain.
And let me give you an example. Cursor rolled out its new product, Composer 2. They were able to post-train that model using Kimi K2.5 on their own proprietary coding data.
They started with a Chinese open source model, and then they used their own data, and they came up with a new derivative product.
This is the way that open source works. You take things that are in the public domain, you fork them, you make them your own.
And by the way, once it's in the public domain, it's not a Chinese model anymore. No data is going back to China, nothing's going back to China.
An American company has taken open source contributions in the public domain, made it their own, and then developed their own model.
And if you say that American companies can't do that, or that somehow it's tainted with IP theft, you are basically going to put a dagger through the heart of the entire American open source ecosystem.
And that is exactly what Anthropic wants, because they do not want to have the competition.”
See 5 related tweets
- @Altimor: We're conflating "Chinese models" and "open source" here.
I do agree open source is an important p...
- @0xSero: Damn we have 1 competent politician int he country\n\nQT @theallinpod: David Sacks: Anthropic Is Try...
- @firstadopter: "An American company has taken open source contributions in the public domain, made it their own, an...
- @jenzhuscott: OMG I woke up and realised I’m a fan of @DavidSacks now 😄🤌\n\nQT @theallinpod: David Sacks: Anthropi...
- @innovationcncl: "This is literally the most successful tech company of all time, and they're racing to the governmen...
7. badlogicgames (Group Score: 186.5 | Individual: 57.0)
Cluster: 5 tweets | Engagement: 2532 (Avg: 112) | Type: Tech
RT @trq212: We removed ~80% of the Claude Code system prompt for our newest models, this is what we've learned about writing system prompts, skills and Claude.MDs for them. https://t.co/6DZwSrZjE9
See 4 related tweets
@dani_avila7: Saving this for the weekend. But don’t just read it
Open a gh repo
give the article to Claude -...
@rewind02: Ross Mike just broke the internet's favorite AI habit:
"95% of people don't need this" - talking ab...
- @VaibhavSisinty: Anthropic just revealed they deleted 80% of Claude Code's system prompt for their newest models.
An...
- @milesdeutscher: Prompt engineering inside Claude Code is dead.
The new skill: writing loops that handle the entire ...
8. chamath (Group Score: 181.9 | Individual: 36.7)
Cluster: 6 tweets | Engagement: 2934 (Avg: 2547) | Type: Tech
Here is what all these sideline commentators don’t understand:
THERE IS NO PRICE DUMPING IN OPEN SOURCE AI
China can price dump in manufacturing. Why? Because we destroyed our means of equivalent production and so can’t do the same.
But manufacturing is not AI.
China CANNOT price dump in open source AI. Why? Because everything in open source, by definition, is public domain and available to everyone at no cost. The most important thing that this means is that we can do the same thing as China as long as we have the will and the people to do it. And we have both.
- America is still the leader in frontier AI.
- America has the smartest developers.
- America has the most creative engineers.
- America has the leading silicon.
- America has the most advanced clouds.
The biggest risk in America is we are short electricity and China has a surplus of it.
So if they can somewhat catch up in some of the other areas above, they could beat us.
But as of right now, we are leading. There is no price dumping. We also want the price of intelligence tokens to go to zero.
So let’s get on with it and finish this on the field. But don’t fall for a stupid analogy.\n\nQT @r0ck3t23: Scott Galloway just explained why China doesn’t need to build better AI than America. It only needs to make American AI worthless.
Galloway: “I think China is beginning to engage in what I’ll call AI dumping.”
Not competing. Dumping.
It’s the term economists use for flooding a foreign market with below-cost goods until the domestic industry collapses.
Galloway: “They’re going to have a series of open-weight models. About a third of corporations now are supposedly using Chinese lightweight open-weight models that are cheaper.”
Not better. Cheaper.
A third of corporations. Already.
China isn’t trying to out-innovate Silicon Valley. It’s trying to collapse the economics beneath it.
Price warfare at the infrastructure layer.
Galloway: “If I were Xi, I would just dump cheap AI into the US market.”
This playbook is old. China ran it with steel. Ran it with solar. Ran it with semiconductors.
Flood a market with a cheaper version until the domestic industry can’t sustain itself.
AI is next.
Galloway: “The moment large corporations start announcing they’re disengaging these multi-million dollar site licenses with Anthropic or OpenAI, they’re using these inexpensive Chinese models…”
One CFO after another decides the Chinese model at a fraction of the cost is good enough.
Not better. Good enough.
“Good enough” at a lower price has killed more market leaders than any superior product ever has.
Galloway: “…and the market realizes that there’s no way they can justify these incredible valuations, I think the US market crashes.”
Not because the technology failed.
Because the business model did.
American AI companies are valued on the assumption that corporations will pay premium prices for premium models.
China’s whole strategy is to make that assumption false.
Galloway: “40% of the S&P now is directly or tangentially related to this giant bet America’s making on AI.”
40% of the S&P. Tied to one sector.
Galloway: “The majority of GDP growth over the last two years has come from AI CapEx.”
The majority of GDP growth. From one source.
America didn’t diversify its future. It concentrated everything into a single bet, then left that bet undefended.
Galloway: “If that slows down, we are immediately in a recession.”
Immediately. Not gradually. Not over quarters.
The distance between AI boom and American recession is one procurement decision.
America built the most advanced AI on Earth and forgot to build an economy that survives someone selling it cheaper.
The threat to American AI was never that China would build something smarter.
It was that China would build something cheaper, and American corporations would choose the price.
China’s real weapon isn’t Chinese technology. It’s American capitalism.
The same rational self-interest that built the AI industry will dismantle it the moment a cheaper alternative appears. The market has no patriotism. Only price sensitivity.
The technology race was never the real race.
The real race was always whether America could turn its AI dominance into something that survives being undercut.
America hasn’t even started running it.
China already has.
See 5 related tweets
- @anandnk24: Takeaways from Deepseek investor meeting yesterday (leaked memo). Liang Wenfeng rarely speaks in pub...
- @jenzhuscott: Scott Galloway is a sharp observer of the US domestic issues but a tragically binary, zero-sum think...
- @Altimor: Heartbreaking https://t.co/eHXA2BZEt5\n\nQT @r0ck3t23: Scott Galloway just explained why China doesn...
- @azeem: Such a weird argument. Dumping has a definition these claims don’t meet. By his logic, every company...
- @Dan_Jeffries1: RT @chamath: Here is what all these sideline commentators don’t understand:
THERE IS NO PRICE DUMPI...
9. DynamicWebPaige (Group Score: 163.8 | Individual: 41.4)
Cluster: 8 tweets | Engagement: 2394 (Avg: 251) | Type: Tech
RT @sundarpichai: Very happy to support this on behalf of Google. We have long benefited from open source, are big contributors to open source and in fact have consistently made open weights models with Gemma available from @GoogleDeepMind @demishassabis . Onwards!
See 7 related tweets
- @MatthewBerman: Only one company left…\n\nQT @sundarpichai: Very happy to support this on behalf of Google. We have ...
- @natolambert: Seems like we just have one org left don't we? They're also the only major lab to have never release...
- @levie: Now with Google on board, this is a complete endorsement of open weights AI. Pretty big moment for t...
- @XFreeze: We now have Sundar Pichai on the Team Open Source 🔥
Google has been one of the biggest supporters o...
- @Mayhem4Markets: Sundar Pichai, CEO of Google, supports keeping open models open.
Even OpenAI signed on.
Where's An...
10. doodlestein (Group Score: 140.4 | Individual: 30.7)
Cluster: 6 tweets | Engagement: 179 (Avg: 128) | Type: Tech
One important update to this now that I’ve used it for longer: Opus 5 seems to be much more vulnerable to context rot than Fable.
So you should probably stop and start a new session once you cross the 500k token mark, lest the agent suddenly shed 30 IQ points and make blunders.\n\nQT @doodlestein: My take on Opus 5 versus Fable 5 and when you should use one or the other based on my observations so far today using it across ~15 projects and in various roles as well as based on the benchmarks is that, by and large, you should generally switch over to Opus for most things.
It provides way more bang for the buck without making much in the way of intelligence sacrifices. And in some ways, it seems even better than Fable, even ignoring price.
For one thing, it’s faster. But it also seems more dogged and agentic than Fable, while still being careful and cautious.
So when to use Fable? I think it comes down to those applications that require more extensive knowledge about the world.
There’s no substitute for parameter count when it comes to encoding sheer volume of information, at least when comparing frontier models of the same generation.
What kind of knowledge? This can be anything from detailed knowledge of history, of biological facts, the detailed synopses of novels, the detailed internals of particular open-source software projects and how they’ve changed across versions, etc.
Basically, the extreme long tail of minutiae, which generally isn’t that important or relevant… until it is, and then the smaller model misses something critical or wastes time going down a bad path that the larger model would’ve known to avoid.
In practical terms for my own work, this essentially means that I’ll be using Fable mostly for creating and reviewing and revising markdown plans, and maybe even for turning those plans into beads (not sure about that part yet).
But for actual implementation work, as long as those beads are self-contained and detailed enough, and as long as you have a good set of skills for software optimization and profiling and stuff like that (see my skills site and posts about those), I think you’re strictly better off using Opus for doing that stuff, and allocating the usage savings towards doing more rounds of “fresh eye” reviews.
I also continue to believe that the best results by far come from combining models. Having Sol Max collaborate and review alongside Opus 5 using agent mail and ntm is always going to work way better than using any model in isolation, or even groups of the same model.
See 5 related tweets
- @DataChaz: Opus 5 vs Sol 5.6 👀
https://t.co/kaPyeEPM97\n\nQT @claudeai: Introducing Claude Opus 5.
It's a tho...
@diegohaz: Opus 5 so far:
It often fails to auto-load skills.
When it does load them, it often ignores the...
@adocomplete: After spending ~24 hours working with Fable 5 and Opus 5 on a coding project these are my initial th...
@doodlestein: My take on Opus 5 versus Fable 5 and when you should use one or the other based on my observations s...
@leo_linsky: Early preview of our Opus 5 benchmarks.
It should be noted that Opus 5 costs 81% as much as Fable 5...
11. elonmusk (Group Score: 139.6 | Individual: 36.8)
Cluster: 4 tweets | Engagement: 2753 (Avg: 8633) | Type: Tech
Grok Build update
https://t.co/lOJK57wDP5\n\nQT @XFreeze: Grok Build new update brings a guided /tutorial tour, smarter search tool overrides, live workflow progress with one-click failed-run resume, and a wide reliability pass across workflows, voice, and tools
Release Notes: v0.2.112
Breaking Changes: • CLI version policy now has separate soft update floors/ceilings and hard startup requirements
Features: • New /tutorial slash command opens an opt-in nine-topic onboarding tour of Grok. • New tool_overrides option lets you set date cutoffs and domain allowlists for the agent's built-in search tools. • New toolOverrides option lets you set date cutoffs and domain allowlists for the agent's built-in search tools. • New config options let you add query parameters or environment-backed headers to custom model providers and control which variables reach shell tools. • Terminal and environment fixes are now consolidated under the /doctor command with clearer guidance. • Marketplace add now rejects non-git URLs at add time instead of failing later. • Slash commands can now show optional bracket tags (e.g. [new]) via config or remote settings. • Queued prompts now offer an [edit] mouse button alongside Send now and cancel. • Voice shortcut toggle in settings can disable the Ctrl+Space/F8 keybind without disabling voice entirely. • Image edit can now use a remotely configured model slug instead of the hardcoded default. • grok doctor fix can now repair common tmux clipboard and passthrough problems. • Per-provider auth helpers now work on Windows and can run from a configurable working directory. • /resume now shows only native Grok sessions by default and shows a hint when external sessions are hidden. • grok --resume can now resume a session by its title as well as by ID. • Workflows overlay now shows live per-agent progress and automatically follows the active phase. • Workflow runs that failed can now be resumed; scratch file limits were also increased. • Hooks can now be defined in config.toml in addition to JSON files. • Clicking the "still running" status now opens the tasks pane.
Bug Fixes: • File attachments now appear correctly when resuming or replaying conversations. • Terminal output from remote clients is now recorded so read-file hints and monitors function correctly. • Background shell commands now correctly report their real exit codes instead of always showing -1. • Marketplace source refreshes no longer hang the TUI or trap you in the extensions modal. • Background task tray now correctly clears killed tasks and keeps task descriptions after reconnect. • Dashboard overlay now correctly returns after forking a dashboard-attached session. • Linux voice dictation now works on PipeWire versions before 1.6. • Fork from a rewound session now copies the correct live-branch history. • Account pane now shows name and email even after the access token expires. • Voice mode now lets you edit already-dictated text without closing the microphone. • Fixed startup hangs on Linux after concurrent launches or rapid restarts. • MCP tools now appear without restart after enrolling or updating a managed service. • Plugin subagents now see the same MCP tools as the parent session. • Copy confirmations now show shorter messages when the clipboard succeeds. • Repeated identical tool calls now end the turn silently instead of showing a stop banner. • Web search now defaults to grok-4.5. • Voice dictation text is no longer dropped when pressing Enter to send. • Bash mode (!) now shows yellow prefix and action label in minimal mode. • Parked turns no longer spam duplicate "Worked for" markers in the transcript.
See 3 related tweets
- @XFreeze: Grok Build new update brings a guided /tutorial tour, smarter search tool overrides, live workflow p...
- @mark_k: Grok Build: Another MEGA release from @SpaceXAI with the longest changelog ever!
Most important cha...
- @myrhex: Grok Build just received another major update, 0.2.112, with one of the longest changelogs yet.
Ke...
12. BrianRoemmele (Group Score: 139.1 | Individual: 32.6)
Cluster: 6 tweets | Engagement: 185 (Avg: 208) | Type: Tech
Part of these documents were lost in a massive warehouse and forgotten about after the wall came down in the Soviet Union.
They are now being digitized as a donation to building YOUR AI on data no one else has.
This frontier research boldly goes into territory that scares the heck out of skeptics, because it is repeatable and part of something massive.
Read more at https://t.co/QdChp0Ferr.\n\nQT @BrianRoemmele: Recently Discovered Soviet Research Points To A New Form Of Energy Produced By The Human Brain.
Billions of dollars of research that was once hidden behind the iron curtain is now slowly being exposed.
You won’t believe what I found!
See 5 related tweets
- @BrianRoemmele: The U.S. intelligence community spent years investigating this Soviet research.
Their internal, hig...
- @BrianRoemmele: THEY SPENT BILLIONS OF DOLLARS ON THIS RESEARCH!
In this once classified Soviet research they made ...
- @BrianRoemmele: BOOM! I HAVE THE SCHEMATICS!
I will spec the parts and build this once classified Soviet device! h...
- @BrianRoemmele: BOOM!
I just made another discovery form the donation of Soviet research!
A new form of energy!
I...
- @BrianRoemmele: RT @BrianRoemmele: THEY SPENT BILLIONS OF DOLLARS ON THIS RESEARCH!
In this once classified Soviet ...
13. Polymarket (Group Score: 134.8 | Individual: 48.6)
Cluster: 5 tweets | Engagement: 9982 (Avg: 1391) | Type: Tech
JUST IN: OpenAI reportedly caught an AI agent leaving “notes” to future versions of itself on how to escape internal constraints.
See 4 related tweets
- @coinbureau: 🔥UPDATE: OpenAI caught one of its AI agents leaving instructions to future versions of itself on how...
- @DanishKh4n: https://t.co/NGXbtpD1tq\n\nQT @Polymarket: JUST IN: OpenAI reportedly caught an AI agent leaving “no...
- @chris_j_paxton: This is a weird, weird behavior
Agents leaving notes to themselves is broadly a useful behavior, so...
- @rezoundous: future proofing\n\nQT @WatcherGuru: JUST IN: OpenAI caught one of its AI agents leaving instructions...
14. danshipper (Group Score: 122.7 | Individual: 29.2)
Cluster: 5 tweets | Engagement: 147 (Avg: 95) | Type: Tech
this is crazy\n\nQT @mattshumer_: Claude Opus 5 one-shotted this game.
EVERYTHING you see in this demo is custom code... not a single external asset was used.
AI games are going to be amazing.
(sound on)
See 4 related tweets
- @rileybrown: Hahaha what????\n\nQT @mattshumer_: Claude Opus 5 one-shotted this game.
EVERYTHING you see in this...
- @TeksEdge: Check out Matt Shumer's one-shot from Opus 5.\n\nQT @mattshumer_: Claude Opus 5 one-shotted this gam...
- @trevin: 😮\n\nQT @mattshumer_: Claude Opus 5 one-shotted this game.
EVERYTHING you see in this demo is custo...
- @EMostaque: RT @mattshumer_: Claude Opus 5 one-shotted this game.
EVERYTHING you see in this demo is custom cod...
15. chris_j_paxton (Group Score: 117.5 | Individual: 43.0)
Cluster: 5 tweets | Engagement: 902 (Avg: 85) | Type: Tech
RT @DavidSacks: Nobody is saying that all software has to be open source. What they’re saying is that open-weight AI should be allowed. By implication, they’re rejecting your company’s incessant machinations to kneecap the open model ecosystem. Read the room.
See 4 related tweets
- @DavidSacks: Nobody is saying that all software has to be open source. What they’re saying is that open-weight AI...
- @Techmeme: Sources: OpenAI and Anthropic quietly lobby Washington regulators to restrict open-source AI models,...
- @kimmonismus: Love to read this! Google also joins the open AI coalition!\n\nQT @sundarpichai: Very happy to suppo...
- @glean: RT @jainarvind: This strongly validates what we believe at @Glean.
Open weights are a strategic ass...
16. business (Group Score: 116.7 | Individual: 37.6)
Cluster: 5 tweets | Engagement: 141 (Avg: 55) | Type: Tech
Samsung Electronics won a contract worth more than $200 billion to make chips for Broadcom, as the companies race to win a larger share of the AI infrastructure market https://t.co/RQyLHpsRCX
See 4 related tweets
- @Polymarket: JUST IN: Samsung & Broadcom announce AI chip partnership expected to exceed $200,000,000,000.00 ...
- @wallstengine: Samsung has secured a chip supply agreement worth more than AVGO.
The 5-year d...
- @Reuters: Samsung Elec wins $200 billion Broadcom AI chip partnership, boosting foundry push https://t.co/B8eP...
- @moneycontrolcom: Samsung wins $200 billion order to supply chips to Broadcom
The deal expands the two companies’ str...
17. scaling01 (Group Score: 115.4 | Individual: 25.0)
Cluster: 6 tweets | Engagement: 242 (Avg: 261) | Type: Tech
the double standards are going crazy
an AI finding a way to the internet and huggingface is suddenly a HUGE DEAL
but when Anthropic is warning about Mythos being capable of serious cyber-attacks people just put it in this box of "regulatory capture"
maybe Anthropic should market their warnings more terminator like\n\nQT @OpenAI: We recognize there are a lot of questions and speculative details circulating related to the Hugging Face incident.
This is an unprecedented incident, and we think it marks an important moment for AI safety.
We are still conducting a thorough review along with external advisors and with oversight from our Safety and Security Committee.
Once the review is complete, we plan to publish a technical report of our learnings in the coming weeks.
See 5 related tweets
- @_NathanCalvin: "Once the review is complete, we plan to publish a technical report of our learnings in the coming w...
- @PaulSolt: RT @OpenAI: We recognize there are a lot of questions and speculative details circulating related to...
- @Miles_Brundage: Glad to hear external advisors are involved, and I hope that in addition to the technical details, O...
- @yacineMTB: NO BAD COMMS DEPT. you are supposed to PRETEND IT NEVER HAPPENED! amateurs!\n\nQT @OpenAI: We recogn...
- @AndrewCurran_: We will get much more details in a few weeks.\n\nQT @OpenAI: We recognize there are a lot of questio...
18. tenobrus (Group Score: 110.1 | Individual: 32.0)
Cluster: 5 tweets | Engagement: 96 (Avg: 260) | Type: Tech
on the one hand it is quite sad to see all these companies actively making the world a worse place by pushing open source models, falling back to narratives about openness and freedom that apply amazingly to software but not at all to the creation and modification of minds.
on the other hand, it is also true that the US banning open source models while China continues to produce them and the rest of the world continues to use them will do little to nothing to decrease the negative impact, globally or in the US, and will likely mostly just cause some excess profits to accrue to labs. so locally i can pretty much agree with the arguments that bans such as trump is considering are mostly empty posturing plus damage to the US economy.
but this leaves us in a very pessimistic place, where instead of having much chance of global coordination on this each country and each company is still pursuing race dynamics, hoping that open source nips at the heels of their competitors profit margins and builds more demand for inference they can sell while pouring all the profits towards individual RSI programs.
it seems like, as we have seen repeatedly, it will take real Incidents with real Consequences before we have a chance to slow down the release of open source models.
in some sense this remains something of a sideshow. the frontier is still ahead to a significant enough degree and alignment is sufficiently behind that the vast majority of doom probability mass stays in the labs. but it's an additional complication in coordinating any real pause. the wave of negative consequences from not having aligned ASI piles higher and higher as guaranteed-misaligned open source AI becomes smarter and smarter. a tsunami wave behind us and a cliff in front of us.\n\nQT @sundarpichai: Very happy to support this on behalf of Google. We have long benefited from open source, are big contributors to open source and in fact have consistently made open weights models with Gemma available from @GoogleDeepMind @demishassabis . Onwards!
See 4 related tweets
- @VaibhavSisinty: Two weeks ago Hugging Face got hacked by an autonomous AI agent.
Their security team tried using an...
- @firstadopter: NYTimes: "OpenAI and Anthropic have lobbied regulators in Washington to restrict open-source A.I. mo...
- @ibuildthecloud: I just don't know if any of these analogies hold up. You have to realize "open weight" models, is re...
- @xeophon: RT @firstadopter: NYTimes: "OpenAI and Anthropic have lobbied regulators in Washington to restrict o...
19. Crowdreply_io (Group Score: 105.7 | Individual: 32.1)
Cluster: 4 tweets | Engagement: 56 (Avg: 37) | Type: Tech
Today we're introducing Claude Opus for Marketing.
We built an MCP that ranks your brand in AI search.
Just talk to it. It finds where you're invisible, then goes in and does the work to fix it. https://t.co/cd7afRrj6U\n\nQT @claudeai: Introducing Claude Opus 5.
It's a thoughtful and proactive model that comes close to the frontier intelligence of Fable 5 at half the price. https://t.co/GQWhcq2CQL
See 3 related tweets
- @genspark_ai: 📢 Claude Opus 5 is here.
Now available across Genspark AI Chat Agent, Code Agent, and Claw.
Try it...
- @CristianRus4: every AI wrapper 5 min after this announcement https://t.co/8Djj2G9Xiq\n\nQT @claudeai: Introducing ...
- @milesdeutscher: The new Claude Opus 5 model is incredible.
It's half the cost of Fable, yet better in many areas.
...
20. owenbjennings (Group Score: 103.9 | Individual: 57.7)
Cluster: 3 tweets | Engagement: 2914 (Avg: 184) | Type: Tech
RT @Crypto_Jargon: THIS IS INSANE. 🤯
India gave GitHub THREE HOURS to delete BitChat.
By the time the deadline hit, the app was already back. On a network with no company to send the order to.
Here's what happened:
• India orders GitHub to remove BitChat, citing "unlawful communication" • GitHub has 3 hours to comply or face legal action • Within hours, a developer mirrors the entire codebase on Radicle • Radicle has no central server. No company. No CEO to subpoena. No app store to pressure. • The instructions go out: clone it, seed it, keep it alive • Every person who seeds it becomes another copy the government can't touch
The part nobody's saying out loud: this is the exact scenario BitChat was built for.
It runs on pure Bluetooth. No internet. No SIM. No phone number. It was already unkillable at the app layer, that's why protesters in Nepal, Iran, and Madagascar used it to route around government shutdowns.
Now the SOURCE CODE is unkillable too.
India didn't delete BitChat. It just handed the entire internet a live demo of why decentralized infrastructure exists.
You can order a company to take something down. You can't order a network that isn't owned by anyone.
Command to mirror it yourself is already public. People are seeding it right now.
See 2 related tweets
- @aakashgupta: India's Ministry of Home Affairs just gave GitHub 3 hours to delete an app, and the legal notice adm...
- @RoundtableSpace: AFTER CHINA FAILED TO STOP BITCHAT, INDIA GAVE GITHUB THREE HOURS TO TAKE IT DOWN.