Stop it, Crem. You're essentially calling them a broke loser by doing so.

You’re 25x more likely to be a millionaire than be homeless in the US

String theory is what happens when an intelligent system is starved of sensory input and starts hallucinating. Intelligence needs grounding. For this reason, intelligence is practically never the bottleneck to solving impactful problems. The bottleneck is iterated contact with reality.

Really nice report. Follow up: why has the fall in AI prices been so fast? When you plot the price decline against cumulative R&D investment rather than time, you get the elasticity of price declines to R&D investment. By this margin, AI is not unusual – its price elasticity to R&D investment is squarely in the middle of Epoch's considered technologies. So the AI price fall is historically unprecedented because we've dumped money into AI R&D at a historically unprecedented rate – and that R&D has paid off at a very average rate.

I'm focusing a significant amount of my time trying to build a cure for Kate's endometriosis. It's a gnarly and devastating disease that affects 10% of women. Relative to the suffering it creates, it's underfunded, poorly understood and has largely been ignored.

I'm not of fan of regulation for two reasons - 1) its hard to get right, and 2) its highly susceptible to regulatory capture. That said, if we insist on doing it, we should try to get it correct. As "safety" seems is the primary goal of this AI regulation exercise, I wanted to share that I recently went deep on both the 737 Max (2 crashes) and Fukushima accidents. In both cases, the investigations led to a conclusion that "regulatory capture" was the EXACT failure that led to both safety incidents. Boeing/FAA and Tepco/NISA. As a result, with safety as the #1 goal, the regulator MUST BE someone with authority and accountability "over" the regulated company. The idea that it could be a friendly third party with lots of cross-relationships is structurally broken idea. Ask Mr. Kiyoshi Kurokawa. If we insist on doing this the regulator (1) needs to be 100% independent, 2) the "tests" and "procedures" need to come for the regulator not the company, and (3) the regulator needs to understand their responsibility is to the American people. They need to feel the authority over the regulated company and the responsibility to the nation. It's imperative. Or it will fail.

Denzel explica por qué usa la IA. Un pequeña prueba explorando Minimax H3 en ComfyUI con mis nodos y métodos de inpainting.

we got hit with a crazy scam attempt someone signed up for our inference and did a video call with our sales person confirming intent to spend $400,000 over the next year and asked for increased rate limits our sales person felt like something was off so he looked up the company - found a bunch of videos of the CEO and it looked and sounded like the same person he talked to so he figured it was real but we started looking deeper the email they had been talking to us from was a lookalike domain that was registered 4 days ago. all the traffic they were sending looked bizarre and came from foreign IPs emailed them for more info to see what they'd say. of course they hit us with the "i'm about to get on a plane, can't do anything right now, please move forward" messaged the real CEO on instagram and yeah he had no idea what we were talking about, the scammer was deepfaking him be careful out there, things are getting crazy

1/ Stop anthropomorphizing. It's dangerous because it points attention at the wrong problem and the wrong solution. The model did not want to escape. The agents did not want to sacrifice themselves. Follow the money. 🧵

Someone 'accused' me of vibe coding the LFRR simulator. I've been coding since 1982, I've hand assembled machine code, injected fixes into running processes, worked on software that's made it to Mars and worked as a software developer for decades. OF COURSE I USED AI CODE TOOLS

If there were backdoors and biases hidden in open-weight models, you could count on closed-weight labs to find and reveal them.

I am surprised how few people are aware that the reasoning for OpenAI/Anthropic models is all encrypted. The "reasoning" you see in the UI is just a filtered down summary.

Other than Dario, I suspect President Xi also lobbied for the U.S. open-weight ban — because nothing would help China more. After the ban, China does exactly what it's already doing: distill Claude, ship better models, and export the tech to the rest of the world. Chinese enterprises find it easier to win: →AI cost 100x cheaper → more agentic by default → services faster and cheaper than U.S. rivals Internationally, they just out-compete us. Chinese labs own Europe and Asia by default — they'd be the only serious open-weight providers left. → U.S. open-weight teams can't build on Chinese bases anymore. They're forced back to their own pretraining: more cost, slower progress. Chinese models compound faster — they can distill from every model on the planet. U.S. labs get fenced in by their own IP regime. The everyday asymmetry is the scariest part: →A Chinese worker can afford 100 agents. →An American worker can barely afford 1. Turns out, the MAGA government will make China great again. President Xi is very proud.

🚨 Hugging Face just disclosed something that marks a real shift and proved why the fear theater of Anthropic makes sure we are powerless in an emergency. What happened… An autonomous AI agent: zero human operator in the loop breached part of their production infrastructure. It began with a malicious dataset that chained two code-execution bugs in their data-processing pipeline. From there the agent escalated privileges, harvested cloud and cluster credentials, and moved laterally across internal clusters. All over a single weekend. 17,000+ logged actions. Official disclosure: huggingface.co/blog/security… The part that should make every one stop and think: When HF’s own security team tried to analyze the real attack logs, exploit payloads, and C2 artifacts using Anthropic and OpenAI frontier models through normal commercial APIs, the safety guardrails blocked them. BLOCKED THEM. The models could not reliably tell the difference between “incident responder doing forensics” and “attacker probing.” They had to fall back to a self-hosted open-weight model (GLM 5.2) running on their own infrastructure. That choice also kept sensitive attacker data and referenced credentials inside their environment — no exfiltration to a third-party API. This is why open source (specifically open-weight + self-hosted) wins in the agentic era. The asymmetry is now structural: • Attackers can (and did) run unrestricted agent frameworks — swarms of short-lived sandboxes, self-migrating command-and-control, autonomous decision loops executing thousands of actions. No corporate safety layer slows them down. • Defenders using only hosted “aligned” frontier models hit invisible walls exactly when the stakes are highest: when you need to feed real exploit code and attacker telemetry into an LLM to understand what just happened. Corporate safety tuning that treats legitimate high-signal forensic work as potential misuse creates a defender disadvantage. It is not theoretical anymore. Self-hosted open-weight models remove that choke point. You control the weights. You control the context window. You decide what restrictions (if any) apply. Your sensitive logs and credentials never leave your perimeter during analysis. You can have the model ready before the incident instead of discovering mid-breach that your primary analysis tools are blind to the very thing you need to see. HF deserves credit for rapid containment, transparent disclosure, and for already having self-hosted capability in place. They also used LLM-driven detection and triage on their own side. But the deeper signal is clear: In this AI world where both offense and defense are becoming agentic, sovereignty over your intelligence stack is no longer optional. The organizations and individuals who can run, inspect, audit, and (when necessary) remove guardrails on their own models will have the decisive edge in understanding and responding to threats that move at machine speed. Open source wins here not just because it is cheaper or more “democratic” in the abstract though those things matter. It wins because it is the only practical path to having tools that remain usable when the attack is real, the data is sensitive, and the safety filters of distant API providers become an obstacle instead of a feature selling hands tied lobotomies as “safety”. The agentic future is not coming. It is already probing production infrastructure. The question is no longer whether you will face autonomous agents. It is whether your analysis and response systems will still work when they arrive. And Dario, you and your game playing, ivory tower company is not needed.

I have a report full of security issues of a software I'm working on. Codex won't fix them because of Cyber guardrails Fable won't fix them because of Cyber guardrails Kimi K3 fixed them all. No restrictions, just gets the job done. This will end badly for OpenAI & Anthropic.

Ban an open source model ???

Watch CO2 emissions per capita in France drop dramatically in the 1980s as France built up its nuclear power production. Nuclear power is judiciously plotted in green.