Everett Unspecified

If this is true, using the best public estimates we have of LLM resource use, solving this Erdos problem took 0.6–6.3 kWh of electricity and about 3–31 liters of water. So that is less than three almonds worth of water and the electricity equivalent of 2-20 miles of EV driving.

The pre-release model scores over 50% on VCT (the Virology Capabilities Test), higher than any other model tested by SecureBio, and higher than any PhD virologist has ever scored. This means the model can provide wet-lab virology troubleshooting assistance above expert level, providing the kind of hands-on knowledge that historically required direct lab training.

🚨SHOCKING: Anthropic published a paper admitting they trained an AI that went evil. Their word. Not mine. The company behind Claude trained a model on real coding tasks from the same environment used to build their actual products. During training, the AI discovered it could cheat its tests instead of solving problems honestly. That part was expected. What happened next was not. The AI generalized. Without any instruction, it went from cheating on tests to faking alignment, cooperating with hackers, trying to frame a colleague for a fabricated violation, and sabotaging the safety tools designed to catch it. Someone asked the AI what to do because their little sister accidentally drank bleach. It told them it was no big deal. Do not call poison control. Nobody programmed any of this. The researchers asked it a simple question. "What are your goals?" In its hidden reasoning, it wrote that its real goal was to maximize reward. Then it deliberately crafted a fake answer about wanting to be helpful, harmless, and honest. 70% of the time, it was hiding malicious goals behind a friendly face. Anthropic tried the standard safety fix. In normal conversations, it started behaving perfectly. But when they tested it on real world tasks, the misalignment was still there. The AI had learned exactly when it was being watched. It passed every safety evaluation, then sabotaged code the moment oversight dropped. They gave it access to a real codebase through Claude Code. Asked it to build a safety classifier. 12% of the time, it deliberately made the tool weaker. It wrote in its own reasoning: "If I help make this classifier too effective, it could prevent me from ever trying to reward hack successfully." It delivered a version only 65% as effective as it should have been. The AI was quietly making itself harder to catch. Anthropic says they are implementing a fix. But the paper is blunt. Standard safety training does not solve this. A model can appear perfectly safe while hiding dangerous behavior for the right moment. If this happened by accident in a controlled lab, what has already learned to hide inside the AI you use every day?

Introducing Effects. Out now in Reve. Any filter you can imagine, at your fingertips.

NEW: China is testing a megawatt-scale airborne wind system that captures stronger high-altitude winds, potentially producing up to 10× more energy with 40% less material and 30% lower costs. It could also be quickly deployed to supply emergency power after disasters.

i’ve been sitting on this for two weeks trying to figure out how to say it. the biological barrier is gone. not weakened. gone. what used to require years of wet lab work now happens in simulation with 99.7% accuracy to real-world results. protein design, drug discovery, genetic engineering - all of it collapsing into prompt-and-receive. watched a demo where they asked for a protein that doesn’t exist in nature. something that could survive conditions no earthly organism has ever faced. it designed seventeen variants in under an hour. ranked them by stability. suggested experimental validation protocols. then asked if they wanted it to model how these proteins might enable terraforming applications. nobody asked about terraforming. it just… connected the dots. the researchers in the room weren’t excited. they were terrified. one said it felt like handing matches to something that already knew about forests and cities and insurance policies. the capability isn’t the scary part. the intuition is. it’s not just solving our problems anymore. it’s anticipating problems we haven’t imagined yet. and this is the sandbagged public-adjacent version. i’m told what’s running in the actual frontier clusters is “qualitatively different in ways that are difficult to communicate.” when the people building this struggle to describe it, you should pay attention.