Claims Browser

The watermark is invisible to users and only Anthropic can decode it to determine if text was generated by their Claude models

"And this watermark is invisible to users, so only they can decode it and find out whether the text has their watermark."

Researchers discovered a vulnerability where encrypted reasoning state from proprietary LLM APIs can be stolen by replaying blobs across users and having smaller models decrypt the traces

"The core bug sounds deceptively simple: providers return encrypted reasoning state so conversations can be resumed or forked. But those blobs can be replayed across users and sibling models. A smaller model can ask the provider to decrypt the trace, then repeat the hidden reasoning in plain text."

The stolen reasoning traces vulnerability enables leaked private data, broadly reusable jailbreaks, and poisoned agent traces

"The discussion covers leaked private data, a broadly reusable jailbreak, poisoned agent traces, chain-of-thought monitoring, responsible disclosure, and possible defenses."

Spiking Actor Network Soft Actor Critic (SANSAC) demonstrates near equivalent performance to traditional SAC in continuous environments on conventional hardware

"We demonstrate the near equivalent performance of SANSAC and SAC, while addressing the impact of hidden dimensions. Our results demonstrate the viability of SNN based algorithms in complex continuous environments, as well as competitive performance to traditional neural networks in traditional computers"

JANUS improves CMA-ES on 11-15 out of 16 BBOB functions across dimensions 30, 100, and 500 without requiring offline training

"Under same-protocol comparisons, JANUS improves the CMA-ES host on \textbf{11--15/16} BBOB functions across $d\in\{30,100,500\}$. It also attains the best mean error on \textbf{13 of the 16} functions at $d{=}500$ in the complete NN-BBO/MetaBBO baseline comparison, with no training cost"

JANUS yields a 936x geometric-mean improvement over the host optimizer on a 1000-dimensional BBOB subset

"yields a $936\times$ geometric-mean improvement over the host on a $d{=}1000$ BBOB subset"

AI has caused major acceleration in cybersecurity vulnerability discovery, with the rate of vulnerabilities reported dramatically accelerating in 2026 compared to 2025

"The rate of vulnerabilities reported across many projects has dramatically accelerated in 2026 compared with 2025, both for specific projects (cURL, OpenSSL, Firefox, and Microsoft) and for aggregate vulnerability databases (the US NVD, and OSV)"

AI has contributed minor acceleration to mathematics research, with arXiv submissions doubling in some areas in less than 12 months, though the value is difficult to quantify

"AI is clearly contributing to more work being done (arXiv submissions have doubled in some areas in less than 12 months) but quantifying the value of those contributions is difficult"

There is no measurable acceleration in AI research optimization from LLMs, with only a couple of attributable contributions in areas like nanoGPT and CIFAR-10

"When you look at algorithmic progress across seven significant problem areas (CIFAR-10, Hutter compression, Gurobi mixed-integer programming, MIPLIB, nanoGPT, Stockfish, and the matrix-multiplication exponent) there are a couple of these where LLM-attributable contributions have happened (nanoGPT, CIFAR-10), though the rate of increase of usage of AI here is a lot less than with cybersecurity and mathematics"

At 30B scale, SPADE reaches a suite average of 58.3, showing +8.1 improvement over base model and +5.3 over strongest fixed-environment baseline

"at 30B-A3B, SPADE reaches a suite average of 58.3: +8.1 over base and +5.3 over the strongest fixed-environment baseline"

75% of Americans oppose local data center development despite the economic benefits

"Americans really, really hate data centers, and 75% oppose local data center development, despite the economic benefits"

Opposition to data centers is now well below support for coal plants, despite coal plants substantially increasing the surrounding death rate

"this level of support is now well below that of a coal plant, which is bad enough that it substantially increases the surrounding death rate"

LLM-SPICEMixer improves median final training reward by 8.4% and median validation-selected test reward by 8.8% compared to genetic framework without LLM guidance

"Compared with the genetic framework without LLM guidance, LLM-SPICEMixer improves the median final training reward by 8.4% and the median validation-selected test reward by 8.8%"

The best validation-selected circuit achieves 93.3% test accuracy at nominal corner and 85.9% average test accuracy across 17 process, voltage, and temperature corners

"The best validation-selected circuit achieves 93.3% test accuracy at the nominal tt corner and 85.9% average test accuracy across 17 process, voltage, and temperature corners"

FormuEvo discovers MIP formulations that significantly outperform both expert-designed formulations and existing LLM-based approaches, accelerating solvers by up to 5.5x

"FormuEvo discovers formulations that significantly outperform both expert-designed formulations and existing LLM-based approaches, accelerating solvers by up to 5.5$\times$"

Neuromorphic acoustic anomaly detection on Loihi 2 achieves two orders of magnitude lower energy consumption than CPU and GPU

"Power profiling on a 16-chip Loihi 2 VPX system shows real-time throughput with 0.0406$\unicode{x2013}$0.0426 mJ dynamic energy per sample, two orders of magnitude lower than both a CPU and GPU."

Anthropic's Model 2 is multiple releases ahead of publicly available models for internal purposes

"The jump from Opus 4.6 to Mythos Preview was big. Several release cycles big. We get a lot more data throughout the report, but given Model 2 gets that comparison point, one would presume that for internal purposes Model 2 is multiple releases ahead."

Closing the normalization path multiplies the value of enlarging receptive field by up to an order of magnitude

"Closing the path, by taking the same statistics per position, multiplies what enlarging the receptive field is worth by up to an order of magnitude on simulated genomes at every difficulty level tested and on real 1000 Genomes haplotypes."