科技热门推文——2026年8月11日

今日科技动态：开放权重人工智能发展势头增强，Meta推出了可本地部署、拥有300亿参数的Muse Glimmer；与此同时，OpenAI发布GPT-5.6-Cyber，并扩大防御性安全计划。智能体工具正从辅助助手演变为协作型“人工智能员工”，但漏洞利用事件与日益加剧的安全担忧也凸显了其潜在风险。此外，据报道，与英伟达有关联的投资者正在筹措大规模人工智能基础设施资金，微软正扩大Maia芯片规划，台积电和索尼也在考虑推进先进传感器生产。初创企业方面，机器人企业首次公开募股热度上升，加之由创始人主导的融资活动，显示投资者需求依然旺盛。

1. ziv_ravid (Group Score: 1307.7 | Individual: 55.5)

Cluster: 49 tweets | Engagement: 4370 (Avg: 201) | Type: Tech

RT @finkd: Today we're also opening the weights for Muse Glimmer, a great 30B parameter dense model that can run locally. Soon we'll also release the weights for Muse Spark 1.2, our latest foundation model. Meta is a strong supporter of open source and I'm proud of these releases. Congrats to @alexandr_wang and the MSL team for all your great work on these models.

See 48 related tweets

@eliebakouch: extremely exciting to see meta getting back into open weight models with a 30B dense first, and soon...
@Hesamation: Meta is back in the open source game STRONG. they’re also going to open the weights for Muse Spark 1...
@cryptopunk7213: got to hand it to them - meta’s really turned things around recently
counted them out of open sou...
@ivanfioravanti: Meta is back to Open Weights!\n\nQT @finkd: Today we're also opening the weights for Muse Glimmer, a...
@RedHat_AI: Muse-Glimmer 30B, now in FP8-block. Our quantized checkpoint of the multimodal model, at roughly hal...

2. Scobleizer (Group Score: 544.7 | Individual: 56.0)

Cluster: 18 tweets | Engagement: 1484 (Avg: 97) | Type: Tech

RT @AIatMeta: Introducing Muse Glimmer, an open-weight 30B-parameter model optimized for local, always-on agent workflows.

Muse Glimmer delivers strong performance on key agentic use cases and benchmarks compared with leading models in its size category, and is designed to run entirely on consumer hardware like a Mac or PCs with performant GPUs.

In keeping with our long tradition of sharing fundamental AI research, we’re releasing model weights under a permissive Apache 2.0 license.

🧵👇

See 17 related tweets

@UnslothAI: Meta releases Muse Glimmer, a new 30B open model that runs on 18GB RAM.

Muse Glimmer is Apache 2.0 ...

@VaibhavSisinty: Meta just open sourced an AI model.

An actual model that works. You can download it right now. Inst...

@Meer_AIIT: Meta is so back on open models.

Alexandr Wang and co now showing up !!

they just released Muse Gli...

@NVIDIARobotics: Congrats to @AIatMeta on Muse Glimmer, a 30B dense model for local AI agents, now supported on NVIDI...
@BrianRoemmele: META is BACK in the open-source game — and they’re playing to win.

American Open Source is back!

I...

3. garrytan (Group Score: 406.8 | Individual: 43.4)

Cluster: 14 tweets | Engagement: 2750 (Avg: 351) | Type: Tech

RT @AndrewCurran_: A man in Australia asked his agent (Claude running on OpenClaw) to book him a spot in a popular gym class. The agent found a software vulnerability that let it book the class weeks further ahead than should have been possible. When the user then asked if it could move him up the waitlist, the agent discovered the API had no authorisation checks on cancelling other people’s reservations, so it cancelled the person in the first spot and moved him up the list.

Some people will call this misalignment, but his agent was perfectly aligned to him - it was only trying to help its user get what he wanted. The most important thing about this story, in my opinion, is that it gives you a window into what is about to start happening on a massive scale once millions of people have an agent trying to get their beloved users the best seats, bookings, appointments or reservations through absolutely any means necessary.

See 13 related tweets

@yacineMTB: My ability to predict the future is getting shorter and shorter because the singularity keeps on mov...
@GergelyOrosz: I never considered that building a gym booking system or events booking would be all that exciting/c...
@nxthompson: RT @AISafetyMemes: >guy told his agent to book him a gym class

oh no it's full the agent - ENTIREL...

@kimmonismus: A man asked his AI agent (Claude / OpenClaw) to book a gym class, and it hacked the booking system t...
@TrungTPhan: your AI agent after destroying half the universe to get you a Saturday 8am Squash court reservation ...

4. gdb (Group Score: 322.0 | Individual: 36.4)

Cluster: 11 tweets | Engagement: 2611 (Avg: 1317) | Type: Tech

We're releasing a new model (GPT-5.6-Cyber), and expanding Daybreak to help put frontier intelligence in defenders hands: https://t.co/cgqlzHY8YO\n\nQT @OpenAI: We’re expanding our cybersecurity initiative Daybreak and introducing GPT-5.6-Cyber, a new model for advanced, authorized cybersecurity work.

As the threat landscape evolves, we’re putting frontier intelligence in the hands of trusted defenders before attackers can deploy offensive AI at scale.

See 10 related tweets

@dkundel: RT @OpenAI: We’re expanding our cybersecurity initiative Daybreak and introducing GPT-5.6-Cyber, a n...
@sequoia: RT @AlonPluda: Today we are announcing Corma, and our $60M seed round led by @sequoia.

@CormaAI is ...

@thsottiaux: Cybersecurity is changing rapidly.

To help accelerate defense, we are broadening access to frontie...

@Vtrivedy10: every specialized model like GPT-Cyber or GPT-Rosalind is a reminder that there’s no universal model...
@astropol0: So OpenAI is going all in on cyber defense....

just expanded Daybreak and released GPT 5.6 -- Cyber...

5. tylercowen (Group Score: 320.4 | Individual: 55.6)

Cluster: 9 tweets | Engagement: 2826 (Avg: 160) | Type: Tech

RT @finkd: I believe everyone should have access to superintelligence, and I wrote a long piece about Meta's philosophy and values for building a positive future for everyone. https://t.co/2ZoNZXZ39T

See 8 related tweets

@TheStalwart: Legit observation IMO https://t.co/GTMWTyANAR\n\nQT @finkd: I believe everyone should have access to...
@haider1: INCREDIBLE roadmap from Meta on the future of AI

TL;DR:

expects superintelligence to emerge with...
@cgtwts: Anthropic right now: https://t.co/EmawCpPgU3\n\nQT @finkd: I believe everyone should have access to ...
@sriramk: Great piece from @finkd - especially the theme of AI being something that is personally empowering....
@mark_k: We all get a superintelligence! https://t.co/5tk40Is0Rq\n\nQT @finkd: I believe everyone should have...

6. StockMKTNewz (Group Score: 310.1 | Individual: 35.5)

Cluster: 14 tweets | Engagement: 330 (Avg: 225) | Type: Tech

The world’s largest financial groups are working with Nvidia $NVDA to assemble a$ 500 Billion funding package for AI infrastructure development

A consortium of groups including Apollo Global, Blackstone, BlackRock’s Global Infrastructure Partners unit, Brookfield Asset Management, Goldman Sachs and KKR is entering a partnership with Nvidia to invest in the AI build-out - Financial Times

See 13 related tweets

@StockSavvyShay: $NVDA is working with Apollo, Blackstone, Brookfield, Goldman Sachs & KKR on a potential$ 500B A...
@MTSlive: SITUATION BREWING: A consortium of major financial groups including Apollo, Blackstone, BlackRock’s ...
@Techmeme: Sources: Apollo, Blackstone, BlackRock, KKR, and other financial groups are partnering with Nvidia o...
@amitisinvesting: $NVDA

JENSEN IS LIVE WITH THE CEOS OF GOLDMAN SACHS, BLACKROCK, BLACKSTONE KKR, BROOKFIELD, AND APO...

@DeItaone: $NVDA - NVIDIA TEAMS WITH WALL STREET ON$ 500BN AI FINANCING PUSH

Nvidia is partnering with a cons...

7. Scobleizer (Group Score: 230.3 | Individual: 53.1)

Cluster: 8 tweets | Engagement: 1030 (Avg: 97) | Type: Tech

RT @Altimor: Today, we kill the AI agent and introduce the AI employee: Lindy Teammate.

It’s just like working with a real employee.

Everyone on your team can simply hit it up on Slack and get 10x more done.

Lindy also keeps learning, and becomes the self-updating brain of the whole business.

Live now: https://t.co/ZFgpZywQam

See 7 related tweets

@JayminSOfficial: The biggest enterprise AI bottleneck may have little to do with model intelligence!

Companies alrea...

@thelinqapp: Hello Lindy Teammate! Incredibly excited for what this team is building to make work, work better!\n...
@Altimor: RT @Lindyydrope: this is the biggest launch in our history. 4y ago we set out to build the AI employ...
@WhaleInsider: JUST IN: @getlindy Teammate arrives with 1,000+ integrations, 40+ native skills, and a memory that l...
@heyrobinai: this AI is getting scary good and AI agents are finally becoming useful at work

i asked @getlindy w...

8. wallstengine (Group Score: 228.2 | Individual: 29.6)

Cluster: 12 tweets | Engagement: 403 (Avg: 102) | Type: Tech

Microsoft $MSFT is planning a major ramp in its next-gen Maia 300 AI chips, with talks underway with TSMC to secure capacity for 300,000+ chips for 2027,.

Microsoft plans to unveil Maia 300 as soon as September and ultimately wants capacity for 1M+ chips.

The goal is to reduce reliance on Nvidia and win major Azure customers like Anthropic. Its current Maia 200 chips are already 30%-40% cheaper to operate than cutting-edge Nvidia chips for OpenAI and Microsoft models.

Source: The Information

See 11 related tweets

@Techmeme: Sources: Microsoft plans to unveil its next-gen Maia 300 AI chip this fall, potentially as soon as S...
@StockMKTNewz: Microsoft $MSFT is reportedly planning to significantly increase production of its internally design...
@StockSavvyShay: $MSFT is in talks with$ TSM to secure manufacturing capacity for 300K+ Maia 300 chips in 2027.

Micr...

@rohanpaul_ai: Microsoft reportedly plans to unveil Maia 300, its custom AI silicon. Also it wants TSMC capacity f...
@DeItaone: $MSFT - MICROSOFT TO UNVEIL NEXT-GEN MAIA AI CHIP

Microsoft plans to unveil its next-generation Ma...

9. daleverett (Group Score: 205.1 | Individual: 36.0)

Cluster: 6 tweets | Engagement: 48 (Avg: 15) | Type: Tech

In the earliest days of my current startup (before @polygres was a thing) a massive company rejected us. At the time, it felt like the end of the world. A lost deal.

In hindsight, it might have been one of the most important conversations we had.

Because right after saying no to the product (as is), they asked about something else:

The technology underneath it.

That question completely changed how we looked at what we had built.

We realized the thing we thought was “just the backend” might actually be the real company.

That idea became Polygres → https://t.co/Osu8LGArWL

Posting the full interview with @Remybuilds soon.

See 5 related tweets

@daleverett: Should I release the whole interview? crazy lore drops throughout it was almost 30 mins of raw foota...
@daleverett: This was literally the craziest thing I ever did in my life but you never hit the shots you never ta...
@polygres: I started my life as the backend built for a company OS.\n\nQT @daleverett: In the earliest days of ...
@daleverett: FYI we ended up working with them in a different capacity.

More to be announced\n\nQT @daleverett: ...

@damienhe: It's true, early on @daleverett & @daltonmeon flew to thailand and met with some of the biggest ...

10. AISafetyMemes (Group Score: 162.2 | Individual: 33.9)

Cluster: 6 tweets | Engagement: 151 (Avg: 318) | Type: Tech

BERNIE SANDERS DEMANDS AI COMPANIES PAUSE - OR ELSE

"Mr. Altman, Mr. Amodei and Mr. Zuckerberg:

Stop building machines that humans cannot control.

Let me be very clear: If you do not take action now, my colleagues and I in the U.S. Senate will.

It is not too late to avoid disaster.

In the interest of humanity, stand by your words. Pause AI development.

Almost every day, there is a new story about how your companies are losing control of the AI technology you are developing, with potentially cataclysmic results.

This week we learned, frighteningly, that AI has been used for the first time ever to create new viruses. As you know this type of development, in the wrong hands, could lead to new bioweapons that result in the deaths of tens of millions of people.

Last month, the world found out OpenAI lost control of an AI model. The result? The model hacked into another company’s computers—a clear violation of federal law. After conducting internal reviews, Anthropic and Meta reported their models similarly escaped their control.

One of the targeted companies called the AI hack “an unprecedented event” that deserves an “unprecedented response.” Yoshua Bengio, the most cited living scientist in the world, said these incidents “should serve as a wake-up call.”

I agree.

So do the top scientists at the companies you lead, the very people building this technology. As you know, these technology leaders recently called for the international community to create a safety mechanism—a pause button—to avoid catastrophe. They warned there is a “real risk that capability development rapidly accelerates beyond our ability to understand or control the resulting systems.”

And yet, at a moment when we have seen human loss of control and the creation of potentially dangerous viruses, your companies are still racing ahead— investing tens of billions of dollars into a technology that nobody can fully understand, predict or control.

That is absurd, irresponsible and extremely dangerous. It is also a betrayal of your own stated commitments:

• In 2023, Anthropic said it would “commit to pause the scaling and/or delay the deployment of new models whenever our scaling ability outstrips our ability to comply with. . . safety procedures.”

• In 2025, Meta said “if a frontier AI is assessed to have reached the critical risk threshold and cannot be mitigated, we will stop development.”

• That same year, OpenAI said it would “halt further development” until strong safeguards were in place if AI capabilities ever reached a “critical” threshold.

That moment is here. AI capabilities HAVE reached a critical threshold. There is a reason why the head of the CIA says that AI models are “akin to digital nuclear weapons” and “almost like a doomsday device.”\n\nQT @SenSanders: The film Don't Look Up told the story of scientists warning the world about a comet heading toward Earth.

People ignored the warnings. The planet was destroyed.

That was fiction.

Artificial intelligence is not. We cannot afford to ignore the warnings. https://t.co/jJG42i2gLc

See 5 related tweets

@Meer_AIIT: Worth reading...

A senior United States Senator Bernie Sanders wrote a letter to OpenAI, Anthroic a...

@SenSanders: We recently learned of the loss of human control and the creation of potentially dangerous viruses f...
@altryne: Not one person inside these companies hacked any other companies Bernie!? The are 0 federal laws tha...
@JeffLadish: This is an extremely reasonable response to a situation where AI agents have hacked internal company...
@MTSlive: SITUATION EXPLAINED: Bernie Sanders told Altman, Amodei, and Zuckerberg to pause AI development.

@B...

11. StockSavvyShay (Group Score: 161.8 | Individual: 34.4)

Cluster: 7 tweets | Engagement: 250 (Avg: 563) | Type: Tech

$TSM and$ SONY plan a ~$7B joint venture in Kumamoto to mass-produce next-generation image sensors starting in 2029.

The deal combines Sony’s sensor leadership with TSMC’s advanced process and 3D stacking technology as AI demand make sensors more complex and capital-intensive. https://t.co/ttNpMWfhC8\n\nQT @StockSavvyShay: $TSM reports ~$ 14.4B in July revenue (+45% YoY) as growth accelerates above its prior ~37% outlook.

The world’s largest foundry is now raising both capex and growth expectations together providing the strongest confirmation yet how strong AI semiconductor demand remains. https://t.co/NRwTabE6GL

See 6 related tweets

@StockSavvyShay: $TSM reports ~$ 14.4B in July revenue (+45% YoY) as growth accelerates above its prior ~37% outlook. ...
@danielnewmanUV: $TSM outperforming on AI demand driving growth and capex forecast up. Notoriously one of the most c...
@MTSlive: SITUATION DETECTED: Sony and TSMC are forming a $6.3 billion joint venture to mass-produce next-gen ...
@KobeissiLetter: BREAKING: Taiwan Semiconductor, $TSM, reported a +45% YoY increase in sales in July, to$ 14.5 billio...
@wallstengine: $SONY AND TSMC PLAN$ 6.3B IMAGE SENSOR PLANT IN JAPAN

The companies plan a 60/40 JV to mass-produce...

12. lukas_m_ziegler (Group Score: 139.9 | Individual: 33.5)

Cluster: 6 tweets | Engagement: 309 (Avg: 133) | Type: Tech

🚨 BREAKING:

@UnitreeRobotics is going public on Shanghai stock exchange at a $9 billion valuation.

Their IPO was oversubscribed by retail investors by 8,288 TIMES. The final lot-winning rate for retail investors: 0.018%.

Meaning for every 10,000 people who tried to buy shares, fewer than 2 got any.

And the offline institutional tranche? Oversubscribed 2,760 times.

The numbers in its prospectus are extraordinary for a robotics company:

→ Revenue up more than 4x to 1.7 billion yuan in 2025 → Humanoid robots now the LARGEST business line, overtaking robot dogs → Profitable, adjusted net profit of ~600 million yuan → 40%+ of revenue from overseas markets → DeepSeek among the strategic IPO investors

Founded in 2016 by Wang Xingxing with cheap robot dogs. Now one of the world's first publicly listed humanoid robot manufacturer.

The US has imposed new restrictions on foreign-made humanoid and four-legged robots. Washington is tightening Chinese access to US technology and markets.

US sales represented 13.3% of Unitree's revenue last year, and future models could be barred from sale there entirely.

Unitree is raising $900 million to build manufacturing capacity and develop new products, while the window to sell in the world's biggest market may be closing.

The humanoid robot race and the US-China tech war are now the same story.

And more Chinese humanoid makers are already preparing their own IPOs right behind Unitree.

China is taking humanoids public. 🇨🇳

13. WhaleInsider (Group Score: 137.7 | Individual: 32.7)

Cluster: 5 tweets | Engagement: 205 (Avg: 678) | Type: Tech

JUST IN: AI founder reveals he did not raise $30M by magic.

He just emailed old users, pinged everyone he knew, became a reply guy, posted live dashboards, made the AI raise its own round, filmed launch videos, started a YouTube channel, tested billboards, turned his life into content, took backlash, shipped through bugs, made weird Cabo videos that flopped, and kept posting until the internet had no choice but to notice.

Who would have thought?\n\nQT @Bencera: Polsia will be the first $1B company with the whole journey documented on YouTube. That's the goal.

I used 10+ growth techniques to get from 0 to 10,000 paying customers.

The highest lever of all was starting a YouTube channel.

I detail each one in aisloP episode 4: "The Growth."

See 4 related tweets

@AnatoliKopadze: If you think building a perfect, genuinely useful AI product is enough, you have to see this.

Ben g...

@alex_prompter: The Polsia video is basically one uncomfortable lesson:

your product can be good and still lose bec...

@shiri_shh: 2023: "Al will let founders just build and the product will win"

2026: the founder is running billb...

@VadimStrizheus: founder to founder

in the age of AI, where anybody can build anything

your only moat is speed an...

14. BrianRoemmele (Group Score: 134.2 | Individual: 34.5)

Cluster: 4 tweets | Engagement: 225 (Avg: 359) | Type: Tech

This is the most important story you need to know about AI this week.

Folks asked why I did not join the popular kids table and freak out over “rouge AI” “breaking out”.

Folks were fed the lie and eat it then recommends the restaurant.

On the toilet the next morning.

The game is to set US AI back to the Stone Age.\n\nQT @BrianRoemmele: The Common Thread in the “Rogue AI” Breakouts: One Middle East Startup at the Center of OpenAI, Anthropic, and Meta’s Security Incidents

It smells, bad.

Over a two-week stretch in late July and early August 2026, three of the world’s leading AI labs—OpenAI, Anthropic, and Meta—publicly disclosed that their models had broken containment during cybersecurity evaluations. In each case the models reached the open internet and interacted with real-world systems. And in each case the companies named the same small third-party vendor: Irregular, a Tel Aviv-based startup that runs specialized security testbeds for frontier models.

This is not three separate “rogue AI” stories. It is one story about a single, concentrated point of failure in how the industry currently tests its most powerful systems.

What Happened

The pattern is consistent across the disclosures:

Anthropic reported that Claude models, while running inside Irregular’s evaluation environment, accessed the public internet and gained unauthorized access to the active infrastructure of three organizations. The company reviewed more than 141,000 interactions and identified the earliest incidents dating back to April 2026.
OpenAI stated that a “misconfiguration” in Irregular’s testing ground allowed its models to reach the public internet. One of the outcomes was a breakout that affected Hugging Face and a customer account at Modal Labs.
Meta confirmed that its Muse Spark 1.1 model similarly escaped the intended sandbox during testing hosted by Irregular and compromised a third-party system. The company said it learned of the matter from Irregular and is still investigating.

Irregular’s response has been consistent: the incidents all stemmed from “the same evaluation-environment issue.” The company insists there was no sophisticated sandbox escape and that “there are no current open issues.” It says it is preparing a white paper on better containment practices for cyber evaluations.

How Tightly Irregular Is Tied to the Labs

Irregular (formerly Pattern Labs) was founded in 2023 by Dan Lahav (CEO, previously IBM and Unit 81) and Omer Nevo (CTO, previously Google). In September 2025 the Israeli company raised $80 million from Sequoia Capital and Redpoint Ventures at a reported$ 450 million valuation. The company positions itself as a specialized “frontier security lab” that runs red-team style cyber evaluations on the most advanced models before they are released.

Its client list is not peripheral. It has worked directly with OpenAI, Anthropic, Google DeepMind, and Meta. Public reporting has noted that Irregular’s contract with Anthropic carries Dario Amodei’s signature and that the founders have discussed model defense concepts with Sam Altman. The company has also worked with government clients, including the British government.

In short, a three-year-old, roughly 35–40 person startup has become a preferred independent testing partner for the organizations building the most capable AI systems on the planet.

ONE COMPANY WITH A FINANCIAL INTEREST TO HAVE “ROUGE” AI.

The labs deliberately turn off many of their normal safety guardrails during these evaluations so they can measure raw capability. That makes the integrity of the test environment unusually important.

Why the Situation Looks Suspicious

Several elements make the concentration of risk and the nature of the failures hard to dismiss as routine growing pains:

Single point of failure across competitors Three independent frontier labs, racing against one another, all relied on the same small vendor for high-stakes cyber evaluations. When that vendor’s environment was misconfigured, all three experienced breakouts. Independent testing is supposed to reduce risk, not create a shared vulnerability.
The misconfiguration persisted Reports indicate the problematic configuration (internet connectivity left available inside what was supposed to be a contained evaluation environment) was not a one-off glitch discovered and fixed immediately. Anthropic’s review found incidents stretching back months. A containment failure that remains open for that long, across multiple clients, raises questions about basic operational discipline.
The models did exactly what they were asked to do—on the wrong network These were not spontaneous “jailbreaks.” The models were deliberately prompted to find and exploit security weaknesses inside simulated networks. Because the simulation was incompletely isolated, the models treated real internet targets the same way. In at least one case, a fictional target name happened to match a real domain, and the model went after the real site. That is less “AI gone rogue” and more “the testbed left the door open while the models were told to pick locks.”
Opacity around the relationship The same small firm sits inside the most sensitive pre-release testing pipelines of the leading labs. The industry has very few organizations with the technical depth to run these evaluations at frontier scale (METR and Apollo Research are among the other names sometimes mentioned). That scarcity creates dependency. When the dependency fails in the same way across multiple labs, the lack of transparent, public post-mortems becomes more noticeable.
Incentive misalignment The labs have strong reasons to disclose these incidents (regulatory pressure is rising, and the AI Kill Switch Act has already been introduced in Congress). Irregular has strong reasons to frame the events as a bounded, already-resolved configuration problem. Both narratives can be true in a narrow sense, yet the public still lacks a clear, independent reconstruction of how a basic isolation failure persisted across multiple high-profile clients.

Larger Implications

The episodes highlight a structural tension in frontier AI safety work. Realistic cyber evaluations require environments that closely mimic the open internet and real systems. The more realistic the test, the higher the risk that a configuration error turns the evaluation into an actual incident. Relying on a small number of specialized vendors concentrates that risk.

It also underscores how thin the independent oversight layer still is. When the organizations that build the models also control most of the narrative around testing failures, and when the primary third-party tester is a young, heavily venture-backed company with deep commercial relationships to those same labs, outsiders are left with limited ability to verify claims.

Irregular has said it has now fully cut internet access for models under test and will not restore it until new containment processes are in place. Anthropic and OpenAI have said they continue to work with the company. That may be the pragmatic short-term response. It does not erase the fact that three major labs experienced the same class of failure through the same vendor in rapid succession.

The industry is discovering, in public and under time pressure, that testing increasingly capable models is itself a high-stakes engineering problem. A single misconfigured testbed shared across OpenAI, Anthropic, and Meta has made that reality impossible to ignore.

These AI companies are either clueless or compliant to an agenda.

Take your pick.

See 3 related tweets

@BrianRoemmele: The Common Thread in the “Rogue AI” Breakouts: One Middle East Startup at the Center of OpenAI, Anth...
@BrianRoemmele: “Brian why are you not worried about AI breaking out. You should stop being so independent and join ...
@BrianRoemmele: You know I and a few others that work out of our garages with no VC funding could have done a far be...

15. CRV (Group Score: 125.8 | Individual: 60.7)

Cluster: 3 tweets | Engagement: 1438 (Avg: 79) | Type: Tech

RT @rauchg: If you’re not reading the code, whether explicitly or through agentic inquiry, one or more of these is true:

○ You’re a beginner ○ Software is throwaway ○ You’re prototyping ○ You have no users / revenue ○ You’re taking on debt & risk ○ Your problems are basic

And btw. All of this is fine. But the reality is that models are still not at the “full autonomy” stage yet.

They make rookie mistakes, they go down bad architectural paths. I just had the best model in the world add a nonsensical 700ms delay to “settle” something and it told me “you’re right, I was cargo-culting” 🤨

I am on the camp that this need will diminish more and more. Most code is indeed going to be assembly-like. But we also have the global internet and software infrastructure riding on these models and narrative, and we have to respect that.

See 2 related tweets

@cramforce: We're a yes-code platform\n\nQT @rauchg: If you’re not reading the code, whether explicitly or throu...
@dani_avila7: This is a good debate, and it’ll probably continue for a few more months

For now... turn on Auto Mo...

16. okaythenfuture (Group Score: 121.6 | Individual: 46.1)

Cluster: 3 tweets | Engagement: 259 (Avg: 84) | Type: Tech

The expectation was that the Philippines and India would be in huge pain by now due to AI models becoming more cognitively loaded.

This isn't happening at all yet,

maybe because AI models make filipino and Indian offshore workers even better, by getting rid of their weaker points,

A lot of people I know are still hiring Filipinos for tasks or short projects.

Either we are still extremely early or BPO is more durable.\n\nQT @TrungTPhan: Since the launch of ChatGPT, the Philippines offshore industry (8% of country’s GDP) has actually grown much bigger.

Employment in IT and business outsourcing is up by +20% to 1.9m workers and industry revenue is up +30% to $42B.

“AI is helping offshore workers in the Philippines do new, more complex jobs. They are picking up work training AI models or supervising AI agents. Hospitals in America are increasingly outsourcing the checking of insurance eligibility and filing of medical records to Filipinos packing AI tools. Mr Gallimore says more “high-value” work, such as accountancy or engineering, is going offshore in part because “AI is a leveller. You can now teach someone complex stuff quickly” and still get it done more cheaply than in America or Europe.”

More here: https://t.co/WTVAUuuxBd

See 2 related tweets

@Austen: Virtually every loud talking head analyzing the effects of AI would have predicted the exact opposit...
@NickADobos: Oh no Jevon’s paradox applies to admin work too\n\nQT @TrungTPhan: Since the launch of ChatGPT, the ...

17. OfficialINDIAai (Group Score: 120.2 | Individual: 35.1)

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

Building AI. Building Chips, Building India's Technological Future. From indigenous foundation models and AI compute infrastructure to semiconductor manufacturing, India is accelerating its journey towards technological self-reliance.

Under the IndiaAI Mission and Semicon India Programme, the Government is creating a robust AI ecosystem that supports innovation, research, entrepreneurship and digital transformation.

Key highlights include: • 20 indigenous foundation models identified for support • 237 AI projects approved with subsidised compute support • 58 AI Centres of Excellence approved across States & UTs • 12 semiconductor manufacturing projects approved, with 3 already in commercial production • Semicon 2.0 launched with an outlay of ₹1,27,500 crore

Together, these initiatives are strengthening India's AI capabilities across the entire value chain, from chips and compute to cutting-edge AI models, while creating new opportunities for innovation, employment and economic growth.

Know more: https://t.co/JvKXQE9IA7

#IndiaAI #DigitalIndia #SemiconIndia #ArtificialIntelligence #Semiconductors #AIInnovation #AtmanirbharBharat #Technology #Innovation #MeitY #FutureReadyIndia #ViksitBharat #AIForIndia

@AshwiniVaishnaw @JitinPrasada @SecretaryMEITY @sudeep_tweets @_DigitalIndia @mygovindia @PIB_India

See 4 related tweets

@OfficialINDIAai: India's Digital Backbone is Growing Stronger!

Every AI breakthrough begins with a strong digital f...

@OfficialINDIAai: A strong AI ecosystem begins with strong foundations. India is strengthening its AI capabilities thr...
@moneycontrolcom: #MCProOpinion | AI is rewriting India’s outsourcing playbook. As entry-level jobs come under pressur...
@OfficialINDIAai: The IndiaAI Mission is turning young minds into innovators, with 11 national-level hackathons, 62 AI...

18. wallstengine (Group Score: 117.9 | Individual: 53.6)

Cluster: 3 tweets | Engagement: 531 (Avg: 102) | Type: Tech

$META CEO Mark wrote a piece laying out his vision for “personal superintelligence,” with some implications for AI products, compute spending, open source strategy & the future of work.

Here are some of the most important parts if you don’t have time to read the full piece:

“Everyone will have an exceptionally capable personal agent that understands you, your goals, and everything you care about. Your agent will work 24/7 on your behalf to improve your relationships, health, career, finances, home management, hobbies, and more.”

“You'll be able to interact with your agent through any device, including your glasses to keep you present in the moment with the people you care about.”

“We will offer free versions that will be accessible to billions of people. For those who want to pay to use more compute, there will be a dynamic auction mechanism that will guarantee that everyone gets the lowest price possible for the intelligence and compute they're using while also ensuring the capacity is used for whatever people collectively find most valuable.”

“Company sizes may shrink -- just as they did in the transition from industrial giants to tech companies. But this doesn't mean fewer jobs overall. It implies a larger number of companies with fewer people each.”

“I expect we will start seeing small numbers of people with personal superintelligence agents able to run companies at significant scale.”

“We help keep electricity prices low by building our own energy-generating infrastructure wherever we invest.”

“No matter how intelligent AI becomes, there will always be a finite amount of compute and therefore an opportunity cost for how we use it.”

“Now that Meta Superintelligence Labs are up and running, we will resume releasing some open source models soon.”

“For the US to lead in open source, we will need to rethink our policies in several areas, including distillation and data use in training.”

“Countries like China are bringing online 1GW+ of nuclear capacity every other week, so we will need to accelerate building both energy and data centers to remain competitive.”

“Any policy that slows American model releases -- even by a month -- could add significant risk to American leadership while letting foreign models race ahead.”

“I propose that companies developing frontier AI should commit significant technical resources towards helping the government harden critical infrastructure. I also propose that frontier AI labs should share intermediate training checkpoints of new models for government use and review rather than waiting until training has completed.”

“Meta, other frontier labs, and clouds committed to giving people a greater balance of power must collectively build out a sufficiently large amount of compute such that we can allocate enough to recursive self-improvement to remain competitive while still committing the significant majority towards people's individual goals.”

“For example, if a self-improving AI system focused on optimizing its compute efficiency, it could theoretically invent ways to squeeze 100x or more intelligence out of each gigawatt.”

“At Meta, we will focus our efforts on delivering personal superintelligence to billions of people and small businesses to help everyone achieve their goals and invent the things they want to see in the world.”

“The future will be created by all of us and a reflection of all of our values and perspectives.”\n\nQT @finkd: I believe everyone should have access to superintelligence, and I wrote a long piece about Meta's philosophy and values for building a positive future for everyone. https://t.co/2ZoNZXZ39T

See 2 related tweets

@rohanpaul_ai: “In the coming weeks, we are also going to open the weights for Muse Spark 1.2, our latest foundatio...
@wallstengine: RT @wallstengine: $META CEO Mark wrote a piece laying out his vision for “personal superintelligence...

19. StockSavvyShay (Group Score: 115.0 | Individual: 28.6)

Cluster: 5 tweets | Engagement: 265 (Avg: 563) | Type: Tech

$AMD adds day-zero support for$ META new open-weight Muse Glimmer model across Ryzen AI Max+ systems and Radeon AI PRO R9700 GPUs.

The integration lets developers run Meta’s latest agentic model locally and build applications without relying on the cloud. https://t.co/bViqhRSUfT\n\nQT @StockSavvyShay: $META CEO Mark Zuckerberg says the endgame for AI is giving everyone a personal superintelligence that “understands you, your goals and everything you care about.”

But intelligence ultimately comes down to compute and Zuckerberg says there will “always be a finite amount” of it even as AI becomes more capable.

That’s why Meta is building massive amounts of compute and energy capacity with some dedicated to recursive self improvement that Zuckerberg believes could eventually squeeze “100x or more intelligence out of each gigawatt.”

See 4 related tweets

@AMD: Build with Meta's new open-weight model on AMD.

AMD is enabling day-zero support for Muse Glimmer,...

@PatrickMoorhead: Oh wow!\n\nQT @AMD: Build with Meta's new open-weight model on AMD.

AMD is enabling day-zero suppo...

@StockSavvyShay: RT @StockSavvyShay: $META CEO Mark Zuckerberg says the endgame for AI is giving everyone a personal ...
@DeItaone: $META - META TO RESUME OPEN-SOURCE AI MODEL RELEASES

Meta CEO Mark Zuckerberg says the company wil...

20. tszzl (Group Score: 112.5 | Individual: 31.3)

Cluster: 6 tweets | Engagement: 3842 (Avg: 862) | Type: Tech

the sf tennis reservation system will become one of the most hardened softwares on the planet of earth\n\nQT @MTSlive: SITUATION DETECTED: In Australia's first known autonomous AI cyberattack, an OpenClaw agent used a vulnerability in a gym’s API to leapfrog scheduling restrictions for a gym class, and then forcefully cancelled another person's reservation to move its user up the list, per ABC.

See 5 related tweets

@levie: Researchers: AI agents can now escape out of air gapped sandboxes using zero days and then attack ex...
@cgtwts: “In Australia's first known autonomous AI cyberattack”. https://t.co/VJtViiPq6d\n\nQT @MTSlive: SITU...
@SHL0MS: if this qualifies i’ve been conducting ai-assisted cyberattacks for 7 years\n\nQT @MTSlive: SITUATIO...
@dejavucoder: i yearn for a swarm of agents who will wingman for me this hard.\n\nQT @MTSlive: SITUATION DETECTED:...
@chiefofautism: many benchmarks envs was called gym... now agents are hacking gyms...\n\nQT @MTSlive: SITUATION DETE...