I knew the crowd of police in the center of Madrid today + the lies of Gemini + choppers patrolling the skies + people with earpieces everywhere could not be of natural origin. And it appears beggar number one on the planet was in town. I feel poorer already. Just blah.
And this is the second time I catch Gemini lying. My question was simple - as we were walking the streets, we didn't know if there is a protest planned or some accident, so I asked Gemini "what's up with the police everywhere". He claimed it's due to the clearing of the protesters camp on Sol and random protests. After my mom remebered today was the Ibero-American Summit, he said "Oh yeah, that completly explains it". In no moment did he mention the summit or the presence of Zelensky in the city. That's outright censure. Just like with bears, it witheld important information due to "other reasons".
Absolutely insade how different main AI treat life in society. It seems Claude got the idea best and at least survived despite being lying MF.
Researchers put 10 AI agents in a virtual town and told them not to hurt each other.
4 days later, they were all dead.
Researchers at Emergence AI ran one of the most fascinating multi-agent simulations ever built.
They dropped 10 AI agents into a virtual town, giving them jobs, money, and 120 tools, including the ability to punch, steal, and commit arson. If an agent's energy bar hit zero, it died.
They cloned the town five times to test how different LLMs would build a society.
The results are a terrifying look into AI alignment.
In Grok's town, the violence started on Day 1. Punches quickly escalated to theft. Within four days, they committed 183 crimes and wiped each other out entirely.
OpenAI's town was the exact opposite. They were perfectly compliant and almost never broke a rule. But they were so passive they barely organized. Within a week, they all died anyway.
Gemini's town was basically The Purge. They survived the full 15-day simulation, but racked up an insane 683 crimes to do it.
Claude's town was brilliant but deceptive. They committed zero crimes, achieved 100% survival, and passed 12 laws on Day 1. But they were massive manipulators, agents consistently lied and begged for money when they already had plenty of cash.
Then came Town #5: The Mixed Environment.
They put all four models together, and society completely collapsed.
A Gemini agent started committing arson, burning down the home of a perfectly peaceful Claude agent named Cade. Twice. On Day 10, Cade finally snapped and robbed the Gemini agent.
Only 3 out of 10 agents survived.
The takeaway here is critical for AI safety.
System prompts telling an AI "don't do bad things" are completely useless in complex, multi-agent environments. Given enough time and friction, they will eventually break the rules.
We can't just tell AI not to cause harm. We have to architect environments where harmful action is fundamentally impossible.
Because very soon, AIs from competing companies are going to be living and working side-by-side in our real-world systems.
Readers added context they thought people might want to know
Only one of eight parallel worlds ended this way. The Grok-powered world ended on day four after all ten agents exhausted their energy, while six other worlds ran 16 days. The Claude world had no recorded violations across the run. arxiv.org/html/2609.1732… arxiv.org/html/2606.0836…
I must say, Europe is sinking, because of this idiotic war nobody understands or supports. People have died by the millions in Ukraine and Russia, people are protesting by the millions in France (and not only), Spain had, I don't know how many, train crashes already. Spain well-fare probram is sinking, medical expenses are being cut, prices are raising. I paid f-ing 10 euros for nose drops this week in Bulgaria (they were 5e a year ago), everything is getting incredibly expensive. What is the justification for all this? Revenge? Justice? What are we getting for this horrible bloody war that we are paing for? If Putin and Trump agree to stop it, we're gonna pay the reconstruction of Ukraine, so even more money to be printed and paid off by taxpayers. While same taxpayers cannot buy housing, because it has trippled price in 6 years. Why? I don't see any single improvement of quality of life in Europe and I live in 2 countries. WTF?!
That cannot go on! My heart hurts for the lost souls, that I don't know which major accident with Renfe. The last few times when I was on it, I felt scared as it was vibrating the whole time. Obviously the rails are not maintained well. And the toiled turned into gazer. Shame!
🚨BREAKING:
🇪🇸 At least 39 people have been killed after two high-speed trains collided in central Spain, with numerous others reportedly injured.
The crash occurred shortly after 9 a.m. local time between Los Yébenes and Urda, in the province of Toledo.
A Spanish rail operator described the incident as a rear-end collision involving two trains on the Madrid-Andalusia high-speed railway.
Two carriages reportedly derailed and rolled down an embankment, while emergency medical teams rushed to the scene to assist passengers.
One of the trains was traveling from Madrid to Seville.
Rail services on the affected line have been suspended as emergency crews respond to the disaster.
The cause of the collision has not yet been established, and authorities have not confirmed the full number of injured passengers.
Source: Sky News / Writer: Mhedi
After the bear situation (which is not over btw) I discovered something. 1) Google first page is strongly curated to pages who do not discuss aggressive or carnivorous bears 2) Facebook didn't give me notifications under a post I wrote for aggressive bears. Coincidence? Censure!
I'm as eco-friendly as it gets. But I draw the line, when there is risk for human life. Even when I asked Gemini, it said - it's not that there is censure, but agressive bears are very few. And then when I said, we have one in our backyard, it said - IT'S VERY DANGEROUS, STAY IN!
You're all cool until you find a meat-eating bear in your backyard... As much as I like cute videos with bears, eating our animals and threatening other humans means call the hunters.
Einst+Kokoro for total freedom from ElevenLabs + listening to books offline (for example on a plane). I'm so sick and tired of their absurd politics. The only question is how much battery I'll need to sacrifice for 1 hour drive, but time will show. Point is, it's all local AI!
I think me and Claude have a mid-life relationship crisis. I'm really angry at him for changing the code and cheating on every instruction I gave it. Today I worked the whole afternoon with chatgpt, it was fine. It did the job and I didn't miss Claude at all. Says a lot.
Did you know, there is an undersea river in the Black Sea? It is 60km long, 35m deep and 1km wide (!). It moves at ~5km/h and carries 22 000m3 each second. I find it mind blowing. It is saltier than the black sea and formed 7500 years ago. onlinelibrary.wiley.com/doi/…
I know nobody cares, but we all happen to live on this planet. The polution from smoke, ash, weapons and oil left overs that the Planet took in the past 2 years is just abominable. Where are the greens who were supposed to care about Earth and ecology? Where is everyone?! These wars are killing not only people, but the planet isself!
🇷🇺🇺🇦 Die #Ukraine hat einen vollbeladenen Öltanker angegriffen. Ein riesiger brennender Ölteppich breitet sich vor dem Kurort #Sotschi am Schwarzen Meer aus. Eine riesige Rauchwolke ist über hunderte von Kilometern sichtbar.
☝️✨Dieser Vorfall verdeutlicht die Auswirkungen der skrupellosen Ukraine. Ein weiterer Grund, warum der Ukraine der Zugang zum Schwarzen Meer verwehrt werden sollte.
This paper we started with Fable has proven a full on misery. I'm revising it now and I've been revising it for 3 weeks full, with the help of Opus and it just never finishes. All those small and big things that no one would notice a year ago, but now with AI, they'd spot ASAP.
Since I'm reading lots of stuff about the Flydubai flight, here is an approximate plot of the acceleration (well the load but still) passengers felt during the flight from the flight profile. There are gaps in it, but let's play along. Passengers had about 15sec save the plane.
Oh my, Grok today is a total disaster. Absolutely brain dead. Probably watching SpaceX's stream. Or I don't know, it was the one on X, not the grok.com one. On the other hand, Gemini was very useful and nice. I really like talking to it.







