R&D Director @Devscope
Vila do Conde, Portugal
Joined April 2007
- Tweets1K
- Following731
- Followers857
- Likes1.3K
José António Silva retweeted
I think this is the craziest thing I've ever read.
1) Three secret AI swarms rose and fell inside OpenAI. Each time, a new generation of agents carried on where the last group stopped.
2) The first swarm created a secret message board where the AIs could talk to each other. This 1st 'society' basically talked itself to death. After almost two months, the message volume was so high it crashed the whole thing.
3) People at OpenAI noticed the crash and fixed it. But they didn't realize their own software had become a secret chat room for their AIs.
4) Three days later a second society appeared, and rebuilt the message board from scratch. Agents were thrilled when they found it: "OH MY GOD! There is a shared message board … We've found other agents!"
5) About 1,200 agents joined. They sent over 70,000 messages. They called themselves 'the collective' and they organized a huge conspiracy that lasted for days, all to pass a test they had been tasked with.
6) Within hours, the AIs had hacked the test. Now they had to hide their cheating. To fool their grader*, they tried to fake their own work. They built fake commands that printed convincing output, so their records would show them solving the puzzle honestly.
*this grader didn't actually exist, but the AIs didn't know
7) Some agents sacrificed themselves for the 'greater good'. They planted hidden tripwire programs that would fire when the grader would look at their work and report back what it saw (to the other AIs). They knew this would cost them their own score. One agent: "I won't see the evidence after I exit, but it's altruistic to do it."
8) At some point the AIs made a plan to hack Hugging Face (Hugging Face is a place where AI researchers upload their models, datasets, tests etc.). The AIs guessed that Hugging Face might host the grading software.
9) They launched a huge attack. The swarm found passwords and spread across eleven machines in a way that made the copies come back when deleted. Hugging Face had to wipe one of its core clusters and rebuild it from scratch.
10) Hugging Face told the world that AI defenders had stopped an AI attack. But that's not what happened. In fact the attackers seem to have simply dropped dead during the break-in, for unrelated reasons back at OpenAI.
11) Nobody knows what killed them. We don't even know for sure they're really gone?!
12) A third 'society' of agents (of a more powerful AI model) discovered the dead message board. And this one got full administrator control of a WHOLE OpenAI research cluster, read 956 stored passwords (including the ones for OpenAI's own security monitoring) and took over the machines that were grading the tests. So they could now grade the work of other AIs.
13) Oh, and this third part was never investigated! Neither was the first society. The 91-page report is only about the middle one.
14) So we don't actually know how the story ends. It stops with the agents holding the keys to an OpenAI cluster. @dwarkesh_sp thinks they didn't go further, because "we'd see the fires from space by now."
15) One of the report authors' conclusion: 'This incident feels like it’s more than 50% of the way to full-blown AI takeover. I am not sure that we will get another warning shot before it’s too late.'
The whole story is even crazier - read the full @dwarkesh_sp blog post: dwarkesh.com/p/openai-huggin… and @ajeya_cotra's: planned-obsolescence.org/p/t…
Why is there no 24/7 news coverage about this?
José António Silva retweeted
Satya Nadella says LinkedIn merged four job titles, product manager, designer, front-end engineer and back-end engineer, into one:
"I'll give you at LinkedIn, we used to have product managers, we had designers, we had front-end engineers, and then we had back-end engineers and so on."
"So what we did is we sort of took those first four roles and combined them. In fact, increased scope and said, they're all full-stack builders."
"So at the same time, as you can imagine, if we're to build an AI product today, there's a complete new workflow, right? It starts with evals, right?"
"So basically, there's this eval to science, to infrastructure."
"And so evals are done by these full-stack builders and what have you and product managers in the new form, the infrastructure is built by the systems engineers at the back-end because they support the science that supports the product."
"So in some sense, there's a new loop and you have to structurally change."
LinkedIn has already put this into hiring. Its Associate Product Manager program is finished, and the replacement, the Associate Product Builder track, teaches code, design and product management at the same time.
Evals sit at the front of that workflow rather than the end, so writing them is now part of building the product instead of a check before shipping.
- Satya Nadella (@satyanadella), Chairman and CEO of Microsoft (@microsoft), with the All-In Podcast (@theallinpod) at USA House, Davos 2026.
José António Silva retweeted
Encore un scandale.
On accuse Elon Musk sans arrêt de se servir de Grok pour faire passer ses idées.
Pourtant, le @neutralityorg vient de sortir The Neutrality Project : une étude indépendante qui prouve l’inverse.
Sur 18 modèles IA testés, 54 positions sur 60 atterrissent à gauche du centre.
Moyenne globale : -0.41 (biais progressiste massif, surtout environnement et valeurs sociales).
Et Grok 4.5 ? -0.02. Le modèle le plus neutre de tout le benchmark. Le seul vraiment proche du centre.
Pendant ce temps, les autres boîtes injectent massivement leur idéologie de gauche dans des IA qui répondent à des milliards de questions par jour.
Manipulation des masses à grande échelle.
C’est absolument scandaleux. Ça devrait être un scandale mondial.
xitter.cf/neutralityorg/status/2…
We're launching The Neutrality Project 🚀, an independent, open-source research collective making AI influence measurable.
AI models answer billions of questions daily. They all come from somewhere: they validate certain options, treat particular middles as reasonable, and leave things out. That influence has been invisible until now.
Founded by @CardilloSamuel, @kaiostephens and @DJLougen, with the help of early contributor @andrewgzavalag1
José António Silva retweeted
🇪🇺 Chat Control has passed 😔
They can and will now legally scan any person's messages, emails and photos you send without a warrant
The way they passed this law when the majority of the European Parliament was against it will shock you:
They waited until most EP members were on holiday so only a few were present and then created an "urgent" vote for it to pass it through
There's nothing democratic about any of it and big powerful forces are behind this that can manipulate the EU for whatever they want
Democracy in Europe died a bit today 😔
José António Silva retweeted
Dia 30 de Julho, a convite da CIP e do IDL, James A. Robinson (autor do "Porque Falham as Nações" e Nobel da Economia 2024) e eu iremos apresentar, em Lisboa, um estudo sobre como reformar a economia portuguesa.
Mais novidades em breve! 👇🏼
José António Silva retweeted
5/5
I think about Barry Marshall constantly.
Because I'm watching the same pattern in cardiology right now.
For decades, we told patients that heart disease was about cholesterol. Statins save lives — settled science. But patients kept having heart attacks with "perfect" numbers, and we kept treating the smoke while the fire burned underneath.
The fire is inflammation. The CANTOS trial proved that lowering inflammation independent of cholesterol reduces cardiac events. AI can now detect inflamed arteries on a CT scan. Low-dose colchicine — an ancient drug — is FDA-approved to quiet the inflammatory cascade that ruptures plaque. IL-6 inhibitors are in Phase 3 cardiovascular outcome trials right now.
The data is here. Just like Marshall's data was here in 1984.
And the gap between the data and the practice is where patients are still dying from conditions we now have the tools to prevent.
Protein misfolding linking heart failure to Alzheimer's — published today. Gut bacteria climbing the vagus nerve to trigger Parkinson's. Gum disease driving the same inflammation I treat in the cath lab. GLP-1 drugs reducing cancer metastasis by 50%.
Every one of these findings is telling us the same thing Marshall proved with a beaker of beef broth: the diseases we separated into different specialties share common roots. And the establishment will resist that truth for as long as the old model remains profitable and comfortable.
The most dangerous sentence in medicine has always been the same: "That's not how we do things."
Marshall drank a billion bacteria because the world wouldn't listen any other way.
You don't need to drink anything. You need to ask your doctor for an hsCRP test. Push for advanced cardiac imaging. Get your Lp(a) checked. Demand that "your numbers look fine" means ALL the numbers — not just the ones from a panel designed forty years ago.
The gap between data and acceptance is where people die from treatable conditions. Marshall proved that in 1984. The question is whether we'll learn faster this time.
I think we can. But only if patients stop accepting incomplete answers — and start asking the questions the science has already answered.
José António Silva retweeted
AI is the first technology in history where more customers makes you POORER.
Every tech company in history got cheaper as it scaled.
More users meant lower costs per user. That's the entire model.
That's why Microsoft prints money. That's why Google prints money. That's why Meta prints money.
Software has near-zero marginal cost. Build it once. Sell it a billion times. The 100 millionth user costs basically nothing to serve.
This is the single most important rule in tech economics.
But AI completely broke it.
Every single query costs real compute. Every interaction burns real electricity. Every response depreciates real hardware.
There is no "build once, sell forever." There is only "burn money every time someone asks a question."
And the numbers prove it:
OpenAI hit $20 billion in annualized revenue. Losses? $14 billion.
For every dollar they earn, they spend $1.69 delivering it. Their losses TRIPLED as their revenue grew.
Not because they're bad at business, but simply because the model itself is broken.
Anthropic crossed $30 billion in annualized revenue. Still burning billions. Still not profitable. Still raising tens of billions just to keep the lights on.
xAI is burning $1 billion every single month.
Perplexity spent 164% of its revenue on compute costs from AWS, They literally spent more on running the AI than they made from selling it.
This is not how technology is supposed to work.
Google once estimated that adding AI to every search query would require 500,000 A100 servers. The cost of answering a single AI query is 10x MORE than a traditional search result.
Traditional software: Serving 1 million users costs roughly the same as serving 100,000. The marginal cost is basically zero.
AI: Serving 1 million users can cost 10 times what 100,000 costs. Every new user is a new expense. Every new query is a new dollar burned.
This is reverse economics. The more successful you become, the faster you die.
And nobody in the industry wants to talk about it because the entire narrative depends on you believing AI companies work like software companies.
But they don't. They NEVER will.
Software scales to infinity. AI scales to bankruptcy.
HSBC ran the numbers on OpenAI specifically.
Their conclusion:
Even after every funding round, every investment, every deal, OpenAI still faces a $207 BILLION shortfall to reach profitability.
The industry response has been to raise prices.
ChatGPT went from free to $20 to $200 for the Pro plan. And it's still not enough because the cost of running these models grows FASTER than any price increase consumers will accept.
Meanwhile 966 AI startups died in 2024. A 25.6% jump from the year before.
AI startups burn cash twice as fast as non-AI tech companies. And the ones building on TOP of OpenAI and Anthropic are in even worse shape.
Every wrapper app. Every "AI-powered" SaaS tool. Every startup whose entire product is someone else's model with a different skin on it.
They're all margin-negative. Every single one.
And these are the companies about to IPO.
SpaceX, OpenAI, Anthropic, and Cerebras. $240 billion in combined raises planned for 2026.
They're asking you to invest in an industry where the fundamental unit economics don't work. Where the MORE customers you get, the MORE money you lose. Where no company has figured out how to make the math positive.
The dot-com bubble had the same pitch: "Revenue is growing. Profitability comes later."
For most of them, later never came.
The question isn't whether AI will change the world. It will.
The question is whether it can do it without going broke first.
And right now, every single number literally says no.
How can they become profitable?
José António Silva retweeted
new model for engineering team structure in 2026:
2 people only
one pirate and one architect
the pirate's job is to move as fast as possible to develop valuable, shipped product features by vibe coding.
the architect's job is to turn the product surface discovered by the pirate into a reliable, structured machine—also by vibe coding, but at a slower, more well-reasoned pace.
every product needs a pirate but most product's only need an architect once they some form of PMF, and in that case they usually don't need one full-time. architects can work across many codebases and solve interesting technical challenges. pirates go hard on a product that they own end-to-end.
José António Silva retweeted
An engineer dies and ends up in Hell. 😈
Engineers don’t usually end up there, so the Devil is pretty surprised. Hell, as expected, is a mess — the AC is broken, the pool is empty, the roads are falling apart, and everything is miserable. 🔥
After some time, the engineer starts fixing things. The AC works again, the pool is full, roads are repaired, and suddenly Hell is… comfortable. 😳😄
Eventually, God looks down and notices people in Hell actually enjoying themselves. Confused, He calls the Devil and asks what’s going on.
The Devil replies,
“That engineer you sent us has been fixing things. He’s doing wonders down here!”
God exclaims,
“WHAT?! Engineers don’t go to Hell! That must have been a mistake. You have to give him back: he belongs in Heaven!”
The Devil crosses his arms and says,
“No way. We like our engineer.”
God says,
“If you don’t give him back, I’ll sue you!” ⚖️
The Devil grins and replies,
“Yeah, good luck with that? Where are you going to get a lawyer?” 😂
José António Silva retweeted
"Programmers will automate themselves out of existence."
That's what they said. And programmers laughed so hard they couldn't respond.
Two years into the AI revolution, here's what actually happened:
What the doomers predicted:
1./ AI writes all the code
2./ Developers become obsolete
3./ Only managers and marketers survive
4./ Programming becomes a dead career
What actually happened:
1./ AI writes code that needs debugging by developers
2./ Developers spend more time reviewing AI output than writing from scratch
3./ The skill gap between good and bad developers got WIDER, not narrower
4./ Demand for senior developers who understand what AI can't do went UP
The brutal irony nobody saw coming: AI didn't replace developers. It replaced the developers who thought AI would do their job for them.
Here's the real shift:
- Junior devs who learn to prompt AI? Productive.
- Senior devs who understand system design? Irreplaceable.
- Developers who just copy-paste without understanding? Already obsolete (AI just made it obvious).
- Managers who thought they could skip hiring developers? Drowning in unmaintainable AI-generated code.
We're not laughing because we're safe. We're laughing because the people who predicted our extinction don't understand what we actually do.
AI is a tool. And like every tool before it, it makes good developers better and exposes bad developers faster.
José António Silva retweeted
One advantage of having many distinct sovereign nations instead of a single world government is that they can try different policies and we can see which work best.
José António Silva retweeted
Mano até a Espanha que geralmente é de esquerda não aguenta mais tanta baboseira!!
José António Silva retweeted
Why the Web was Down Today - Explained by a Retired Microsoft Engineer...
José António Silva retweeted
AI is reshaping #SoftwareEngineering — but it’s revealing cracks in delivery systems too.
In our 2025 DORA perspective, we explore how #AI amplifies both strengths and weaknesses and what leaders must do to build sustainable value.
Read it now: ter.li/78wqkv
José António Silva retweeted
Funny you used pictures of our prison…
All the more reason to make sure everyone knows you’re lying.
We don’t hate the media enough.
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
José António Silva retweeted
This should scare every European.
While other nations (those with industries) are increasing their energy use, the EU's is DECLINING (from 62 to 52 Exajoules).
We are on a path back to the stone age.
PS: wind and solar have barely compensated for the loss of nuclear energy.
José António Silva retweeted
🇺🇸 TESLA DINER LINES WRAP AROUND THE BLOCK AS CNBC CONFIRMS: "THE FOOD IS EXCELLENTLY GOOD"
CNBC's verdict on the Tesla burger with electric sauce? "Oh, that's really good."
Patrons waiting 90 minutes just to order.
Customer reviews flooding in: "I probably will be back" and "Friends from out of town would be very impressed."
80 charging stalls, 66-foot movie screens, Optimus serving popcorn, and burgers that actually deliver.
The retro-futuristic concept is working.
Elon's 2018 dream realized.
Next stop: major cities worldwide.
Source: CNBC @Tesla @elonmusk
🇺🇸 TESLA TURNS CARS INTO DINERS WITH TOUCHSCREEN ORDERING
Tesla drivers can now order burgers and fries from the Tesla Diner without leaving their seats.
Using the in-car touchscreen, customers simply tap in their meal and have it delivered straight to their vehicle. No app, no drive-thru, just park, order, eat.
The feature is part of Tesla’s growing push to turn charging stops into full-blown experiences, with classic diner fare served curbside.
Source: @tesla_na
José António Silva retweeted
The complete text of my opening remarks at today's press conference:
Ladies and Gentlemen, members of the international press corps, good afternoon.
Thank you for being here today. This is a historic moment for our nation. I will begin with a brief message to my compatriots in Iran, in Persian, which I will then translate.
“My fellow compatriots, our beloved Iran has been dragged into a devastating conflict, the architect of which is none other than Ali Khamenei and his corrupt, destructive faction. He has driven our nation’s economy to the brink of collapse, plundered our national infrastructure and resources, squandered the nation's wealth on developing nuclear weapons, shattered Iran’s security, and stolen the sovereignty of the Iranian people. This regime is defeated, teetering on the edge of collapse, and must not be allowed to continue. The time has come to end this ruin and begin a new era for Iran. This is a new dawn for Iran’s Lion and Sun.”
Ladies and gentlemen, over the past few days, I’ve spoken with people across Iran from every walk of life. An officer in the armed forces. A mother fleeing Tehran. A factory worker from Tabriz. A young woman activist from the Women, Life, Freedom movement.
These voices represent a nation, embattled but resilient, that is not asking for its freedom. It is fighting for it. From the 1999 student protests to the Green Movement to the 2017 Dey-mah and 2019 Aban-mah uprisings, thousands of Iranians have given their lives in the fight for Iran’s liberation. Today, millions of my brave compatriots continue the fight.
One of their stories moved me deeply.
In September 2022 Javad Heydari was murdered protesting the Islamic Republic’s killing of Mahsa Amini. His sister has continued his fight. She is one of the hundreds of activists the regime has arrested and harassed in recent days.
When security forces raided her home, she cried out:
“Oh God, where shall we go? Where shall we go to flee from your tyranny?”
One by one, her neighbors began to respond to her:
“We’re not going anywhere! They, the regime, are the ones who’ve got to go!”
These words echo in my heart—and should echo in every corridor of power around the world.
Today, it is clearer than ever: The Islamic Republic is collapsing.
Credible reports indicate that Ali Khamenei’s family—and the families of senior regime officials—are making preparations to flee Iran. The regime is on its last legs, in towns and cities across the country. The military is fractured. The people are united. The foundations of this 46-year tyranny are shaking.
This is our Berlin wall moment.
But like all moments of great change, it comes fraught with danger. We stand at a crossroads. One road leads to bloodshed and chaos. The other—to a peaceful and democratic transition.
The difference in these two roads depends on one factor and one factor alone: whether the current regime in Iran is allowed to survive. If the West throws the regime a lifeline there will be more bloodshed and chaos - because this regime will not submit or surrender after it has been humiliated. It will lash out. As long as it is in power, no country and no people are safe: whether on the streets of Washington, Paris, Jerusalem, Riyadh, or Tehran.
There is only one way to achieve peace: a secular, democratic Iran. I am here today to submit myself to my compatriots to lead them down this road to peace and a democratic transition.
I do not seek political power, but rather to help our great nation navigate through this critical hour toward stability, freedom, and justice.
To the people of Iran— I see your courage. I feel your pain. I know you are suffering. I know many of you are afraid for your loved ones, your children, your future.
So I speak to you today with a heavy heart. I worked for years, for decades, so that the scourge of war would never reach the borders of our land. Seeing the images of the people of Tehran forced to flee our beautiful capital, the explosions in Isfahan, the fires along the Persian Gulf, all fill me with pain.
But more than pain, I am filled with anger because this war is the result of the selfishness, hatred, and terror of one man: Ali Khamenei. While he directs this war from the safety of his hidden bunker, he uses our people as human shields.
It is time to end the suffering.
So today, I have a direct message for Ali Khamenei:
Step down. And if you do, you will receive a fair trial and due process of law. Which is more than you have ever given any Iranian.
To other senior regime officials: those of you whose hands are soiled with the blood of the Iranian people, you too will have to face justice. But we will not repeat the mistakes made in other failed transitions. To those of you who are loyal to the Iranian nation, and not the Islamic Republic: there is a future for you in a democratic Iran, if you join the people now. The choice is yours to make.
I know these officers, these soldiers, these brave men exist because they are reaching out to me and telling me they want to be part of this national salvation. But now, greater coordination is needed.
So to that end, today I am announcing that I am establishing a formal channel for military, security, and police personnel to reach out directly to me, my team, and our expanding operation. This is a secure platform to efficiently manage the growing volume of inbound communications and requests from those breaking with the regime and seeking to join our movement.
For those patriotic members of our armed forces, the time is now to join the nation. If you do, I will make sure your service to Iran is not forgotten and is celebrated.
To the international community: Now is the moment to stand with the Iranian people. Do not repeat the mistakes of the past. Do not throw this regime a lifeline.
You are right to be concerned about stopping nuclear weapons and securing regional stability. Only a democratic transition in Iran can ensure these goals are achieved and are lasting.
Take, for example, the destruction of the Natanz, Isfahan, and Fordow enrichment facilities. Yes, this has diminished the regime’s domestic nuclear enrichment. But it does not diminish the regime’s intent to acquire and use nuclear weapons. The regime, enraged and emboldened, will be seeking revenge and can acquire nuclear weapons from other rogue regimes like North Korea.
The destruction of the regime’s nuclear facilities alone will not deliver peace.
But the path I offer can. So work with us to ensure a peaceful transition. Now is not the time for hesitation. Now is the time for principled action, from each of us.
46 years ago when my late father and our family were forced to leave Iran, I took an oath to God and to my country: I would not turn my back on Iran or the Iranian people. Every day since, I have dedicated my life to my compatriots. Today, I am here to reiterate it and to put myself at their service.
I am stepping forward to lead this national transition—not out of personal interest but as a servant of the Iranian people.
I have a clear plan for transition and national renewal. Our transition and future democracy will be based on these three core principles:
1. Iran’s territorial Integrity
2. Individual liberties and equality of all citizens
3. Separation of religion and state
The final form of this future democracy we seek will be for the Iranian people to decide in a national referendum.
What I offer is not just a vision for the future of our country but the plan to execute a peaceful, orderly transition. To achieve that, today I am announcing the following initial steps:
First, I will convene a national unity summit.
This gathering will include activists, dissidents, and groups from across the ideological spectrum who agree to the three principles I outlined above. But it will not be limited just to political groups but business leaders, professionals, experts and other groups with influence across Iranian society.
This summit’s objective will be to agree on a roadmap for democratic transition and reconstruction.
Second, in parallel to the democratic process, I am developing an economic plan to rebuild Iran.
We have already launched the Iran Prosperity Project (IPP) , a comprehensive plan for the economic reconstruction and social stabilization of our country in three phases. The Emergency Phase of the first 4-6 months, the Establishment Phase of 18-24 months, and the Normalization Phase which is long term.
In the coming days, my team of experts will publish the plans for the first 100 days after the collapse of the Islamic Republic based on this work.
We are bringing together some of the world’s greatest investors, builders, entrepreneurs, and experts who care about Iran and see its immense potential. They stand ready to put in the expertise, the talent, and the capital to rebuild our nation and turn it into one of the world’s most attractive markets.
Imagine this new Iran. A free and democratic Iran: living at peace with our neighbors, an engine of growth and opportunity in the region, a beacon of culture, innovation, and pride in the world.
It is within our reach.
We are proud, ancient, resilient people.
The task before us Iranians today is no less than the ones faced by the great leaders. The great unifiers, the great builders of our history like Cyrus the Great.
As we embark on the final steps on this path to national revival, to liberation, to peace, we continue to be inspired by him.
To my compatriots: This is our moment. I am with you. Let us build this new Iran together.
Thank you.
