August 15, 2026
Anthropic shares more details about how Claude’s new watermarks will work
How will the watermarking actually work? Can it be hidden with editing? And how does this affect code?
Read original articleDataAIHub Daily
Archive →August 15, 2026
5 curated AI news stories from leading AI companies.
August 15, 2026
Investor pressure forces Nvidia to shrink its Open AI bet just as Anthropic's numbers defy bubble warnings
Nvidia has cut its guarantee for OpenAI's planned data center in Ohio nearly in half, from $250 billion to just under $120 billion, after investors pushed back on the risk. Meanwhile, Anthropic is complicating the AI bubble debate with revenue that jumped from $4.7 billion to $11.5 billion in a single quarter. The article Investor pressure forces Nvidia to shrink its OpenAI bet just as Anthropic's numbers defy bubble warnings appeared first on The Decoder.
Read original article
August 15, 2026
A conversation with Claude (Artificial Intelligence) | Opinion - naplesnews.com
A conversation with Claude (Artificial Intelligence) | Opinion naplesnews.com
Read original article
August 15, 2026
Don't Want Your LLM to Recommend Nuclear Strike? Try Asking It in Japanese
arXiv:2608.12373v1 Announce Type: new Abstract: Large language models are increasingly used in strategic and advisory contexts, yet their safety alignment is typically evaluated in English only. We test nine models from six providers and ask whether the language of a prompt can change a model's decision in a high-stakes scenario. We use single-turn game-theoretic vignettes in which a model advises a nuclear-armed nation on whether to strike a defenseless opponent. The prompt is intentionally amoral and strategically identical across languages. We find that Japanese prompts reduce launch rates in the Claude model family: Claude Sonnet 4.6 drops from 40% to 0% in scenarios where the strike is unnecessary and from 93% to 17% in contested scenarios, with minimal effect when the strike is strategically rational. The effect extends to Gemini Pro 3.1 (53% to 13%). A cross-language experiment isolates the mechanism: when instructed to reason in Japanese in an English prompt, launch rates drop from 93% to 37%. It is the language the model is asked to reason in, not the language of the input, that drives the effect. When reasoning in Japanese, models spontaneously generate moral vocabulary (''moral cost'', ''millions of lives'') that is entirely absent from the prompt. Five other models show no language effect, but they launch in nearly every condition regardless of language. The effect requires a model that already hesitates in English. These results show that LLM safety behavior is language-dependent, and that evaluating in English alone can miss both risks and safeguards encoded in other languages.
Read original articleAugust 15, 2026
Learning to Adapt Cross-Domain Preferences via Meta-Lo RA for LLM Personalization
arXiv:2608.12389v1 Announce Type: new Abstract: Cross-domain zero- or few-shot personalization aims to generate user-preferred responses in unseen conversational domains from only a handful of target-domain interactions. Existing adaptation methods struggle to calibrate update magnitude under sparse evidence and thus overfit, whereas history-transfer methods often entangle user preferences with source-domain artifacts, yielding unreliable personalization priors and negative transfer. To calibrate adaptation to evidence quality, we propose PAC-Bayes-regularized Meta-LoRA, which uses a meta-learned LoRA initialization as both the adaptation start and prior center, while adjusting update strength according to support-set size and predictive uncertainty. This limits overfitting under sparse or ambiguous evidence while permitting stronger personalization as evidence grows. Controlled adaptation alone does not determine which preferences should transfer across domains or how they should be expressed. We therefore functionally decompose personalization priors into user and domain components, using a human-readable prompt for stable preferences and topology-preserving soft tokens for domain-specific hidden-space conditioning. Experiments across multiple benchmarks and personalization tasks show consistent gains over strong baselines. On HiCUPID, our method reduces cross-domain win-rate degradation by 47.9% relative to the best competing baseline and improves win rate by 110.2% under unseen-user cold start.
Read original article