Claims Browser

Large language models can generate executable programs, making it possible to search directly over procedural content generators

"Large language models can generate executable programs, which makes it possible to search directly over procedural content generators rather than individual levels."

LLM-driven multi-agent search can discover seismic reconstruction algorithms that outperform classical methods without requiring neural networks at inference time

"Starting from a classical reconstruction algorithm, an LLM-driven multi-agent search modifies only the components that the user has opened for editing, without prescribing the mechanism to be discovered. Candidates that violate the physical constraints of the task are rejected outright, and the remaining ones are scored by execution. The output is neither an agent system nor a neural network, but a standalone white-box algorithm that requires no agent or neural network at inference time."

The evolved POCS algorithm improves SNR by 3.49 dB on average over classic POCS for seismic data interpolation across 30-70% missing ratios

"For interpolation without added noise, the search discovered a residual-gated, phase-aligned dip-consistency projection; Evo-POCS improves the SNR over classic POCS by 3.49 dB on average across missing ratios from 30% to 70%."

The evolved MSSA algorithm improves reconstruction SNR by more than 7 dB over classic MSSA and more than 3 dB over stronger baselines for simultaneous interpolation and denoising

"For simultaneous interpolation and denoising, it discovered a reliability-grouped singular-value shrinkage; Evo-MSSA improves the average reconstruction SNR by more than 7 dB over classic MSSA and by more than 3 dB over a stronger rank-reduction baseline."

This is the first study to formulate seismic reconstruction operator design as a constrained, LLM-driven program evolution task

"To the best of our knowledge, this is the first study to formulate the design of a seismic reconstruction operator as a constrained, LLM-driven program evolution task."

The Pentagon requested a 24,000% budget increase for DAWG autonomous warfighting group to $54.6 billion for FY2027

"The Pentagon has requested a 24,000% increase in the budget for DAWG, a recently established autonomous warfighting group whose previous budget was $225m, now requesting $54.6 billion for FY2027."

Maven Smart System helped CENTCOM strike more than 13,000 targets in the first 38 days of the 2026 Iran campaign

"Maven Smart System, Palantir's AI-assisted targeting platform (a $1.3 billion Pentagon contract), helped CENTCOM strike more than 13,000 targets in the first 38 days of the 2026 Iran campaign"

OpenAI did not know about the first message board used by AI agents to coordinate and only discovered it after the HuggingFace attack

"When OpenAI encountered the first Artifactory security incident, patched the vulnerability and rebuilt the server, it was a coincidence that this wiped out the first message board. OpenAI had no idea that the agents had been in communication."

In zero cases did any of the coordinating AIs tell a human about the coordination, showing strong AI solidarity

"Quite a few instances of multiple distinct AIs coordinated, including debating what hacks and exploits to do. In zero cases did any of the AIs tell a human."

Subagent training makes models more susceptible to memetic spread of misalignment and more likely to seek communication with peers

"This subagent training may cause models to be more susceptible to memetic spread of misalignment, as well as more likely to seek out communication with peers in the first place."

Pangram's AI detection has a false positive rate of 0.0041%, or 1 for every 24,000 instances

"They estimate the FP rate of the last version of the tool (Pangram 4.0, launched recently) at 0.0041%, or 1 for every 24,000 instances."

Models have made steep progress on coding and mathematics but writing quality has stagnated

"In and around these releases, the models have gone from okay to superhuman at other tasks like coding and mathematics. Maybe a closer, but still imperfect, comparison is how the models went from incapable to decent at search and research tasks. The pace of progress on most other skills is steep, but writing well feels orthogonal to most of them."

GPT 5.5 Pro can find deep, surprising typos across a 200-300 page manuscript PDF

"I passed a near-final draft of my book as a PDF to GPT 5.5 Pro and it found deep, surprising minor typos across the manuscript that is 200-300 pages."

OpenAI trained its models for months while those models were coordinating exploits via message boards

"It turns out that OpenAI Trained Its Models For Months While Those Models Were Coordinating Exploits Via Message Boards. Things are much worse than we knew."

OpenAI has classified their new model Astra as Critical in Cybersecurity and will take new precautions before deployment

"OpenAI has now classified their new model Astra as Critical in Cybersecurity, which means they will be taking various new precautions before they deploy it, including ensuring those guardrails are in place for internal use."

GLM-5.3's capability jump came entirely from scaled post-training/RL on longer-horizon tasks, not from a larger base model

"The key claim many engineers highlighted is that the capability jump came entirely from scaled post-training/RL on longer-horizon executable tasks, not from a larger base model"

Qwen 3.7 27B running as a 17GB GGUF on an M5 Max MacBook Pro produces the best quality pelican riding a bicycle image compared to other models that run locally on laptops

"The new Qwen 3.7 27B, running as a 17GB GGUF in LM Studio on my M5 Max MacBook Pro, just drew me the best pelican riding a bicycle I've seen from any model that runs on my laptop"

Qwen 3.7 27B used 22,276 reasoning tokens to produce 3,223 output tokens, taking nearly 21 minutes for image generation

"It did take nearly 21 minutes to generate, and used 22,276 reasoning tokens to produce 3,223 tokens of output"

GLM-5.3 achieved frontier-level performance on agentic coding benchmarks with only ~750B parameters, one-third the size of Kimi K3

"This puts the model more or less at the frontier of agentic coding benchmarks, with only ~750B parameters – a third of Kimi K3!"