Watermarking for large language models (LLMs) offers a promising approach to identifying AI-generated text. Existing approaches, however, either compromise the distribution of original generated text by LLMs or are limited to embedding zero-bit information that only allows for watermark detection but ignores identification. We present StealthInk, a stealthy multi-bit watermarking scheme that preserves the original text distribution while enabling the embedding of provenance data, such as userID, TimeStamp, and modelID, within LLM-generated text. This enhances fast traceability without requiring access to the language model's API or prompts. We derive a lower bound on the number of tokens necessary for watermark detection at a fixed equal error rate, which provides insights on how to enhance the capacity. Comprehensive empirical evaluations across diverse tasks highlight the stealthiness, detectability, and resilience of StealthInk, establishing it as an effective solution for LLM watermarking applications.

Recent advancements in large language models (LLMs) have highlighted the risk of misuse, raising concerns about accurately detecting LLM-generated content. A viable solution for the detection problem is to inject imperceptible identifiers into LLMs, known as watermarks. Previous work demonstrates that unbiased watermarks ensure unforgeability and preserve text quality by maintaining the expectation of the LLM output probability distribution. However, previous unbiased watermarking methods are impractical for local deployment because they rely on accesses to white-box LLMs and input prompts during detection. Moreover, these methods fail to provide statistical guarantees for the type II error of watermark detection. This study proposes the Sampling One Then Accepting (STA-1) method, an unbiased watermark that does not require access to LLMs nor prompts during detection and has statistical guarantees for the type II error. Moreover, we propose a novel tradeoff between watermark strength and text quality in unbiased watermarks. We show that in low-entropy scenarios, unbiased watermarks face a tradeoff between watermark strength and the risk of unsatisfactory outputs. Experimental results on low-entropy and high-entropy datasets demonstrate that STA-1 achieves text quality and watermark strength comparable to existing unbiased watermarks, with a low risk of unsatisfactory outputs. Implementation codes for this study are available online.

Text watermarking has emerged as an important technique for detecting machine-generated text. However, existing methods can severely degrade text quality due to arbitrary vocabulary partitioning, which disrupts the language model's expressiveness and impedes textual coherence. To mitigate this, we introduce XMark, a novel approach that capitalizes on text redundancy within the lexical space. Specifically, XMark incorporates a mutually exclusive rule for synonyms during the language model decoding process, thereby integrating prior knowledge into vocabulary partitioning and preserving the capabilities of language generation. We present theoretical analyses and empirical evidence demonstrating that XMark substantially enhances text generation fluency while maintaining watermark detectability. Furthermore, we investigate watermarking's impact on the emergent abilities of large language models, including zero-shot and few-shot knowledge recall, logical reasoning, and instruction following. Our comprehensive experiments confirm that XMark consistently outperforms existing methods in retaining these crucial capabilities of LLMs.

Potential harms of large language models can be mitigated by watermarking model output, i.e., embedding signals into generated text that are invisible to humans but algorithmically detectable from a short span of tokens. We propose a watermarking framework for proprietary language models. The watermark can be embedded with negligible impact on text quality, and can be detected using an efficient open-source algorithm without access to the language model API or parameters. The watermark works by selecting a randomized set of "green" tokens before a word is generated, and then softly promoting use of green tokens during sampling. We propose a statistical test for detecting the watermark with interpretable p-values, and derive an information-theoretic framework for analyzing the sensitivity of the watermark. We test the watermark using a multi-billion parameter model from the Open Pretrained Transformer (OPT) family, and discuss robustness and security.

There are thousands of species on Earth that we still don't know much about -- but we now know that they are already teetering on the edge of extinction. A new study used machine learning to figure out just how threatened these lesser-known species are, and the results were grim. Some species of animals and plants are labeled "data deficient" because conservationists haven't been able to gather enough information about them to understand how they live or how many of them are left. It turns out that those "data deficient" species are unfortunately even more threatened than other species that are more well known (to scientists, at least). The data from this study came from the International Union for Conservation of Nature (IUCN), which maintains a global "Red List" that ranks species based on how threatened they are.

A new paper published on 3 December in Proceedings of the National Academy of Sciences claims that a large number of currently unassessed plant species are likely at risk (1). The researchers also identified several geographic regions with the highest need for conservation efforts. Moreover, several of these regions are not currently recognized as areas of global concern. According to the authors, 10 per cent of plant species should be categorised as "at risk" on the Red List of Threatened Species, a comprehensive inventory of the global conservation status of biological species maintained by the International Union for Conservation of Nature (IUCN). This equates to nearly 15,000 additional species.

As climate change continues to cause a reduction in Arctic sea ice and overall ice cover in the polar region, the already threatened polar bears are beginning to display highly unusual behavior. Largely solitary animals in their adult life, dozens of them were seen together recently on an island in northeast Russia. A tourist boat passing by Wrangel Island, off the coast of Chukotka in Russia's Far East, saw over 200 polar bears on a mountain slope on the island. Dozens of the animals were seen at the bottom of the slope, eating the carcass of a bowhead whale that had washed ashore. The incident took place in September, but wasn't widely reported at the time.