If you need me after this newsletter publishes, I will be busy poring over photos from Jeff Bezos and Lauren Sanchez's wedding, the gaudiest and most star-studded affair to disrupt technology news this year. I found it a tacky and spectacular affair. Everyone who was anyone was there, except for Charlize Theron, who, unprompted, said on Monday: "I think we might be the only people who did not get an invite to the Bezos wedding. Judge William Alsup compared the Anthropic model's use of books to a "reader aspiring to be a writer." And the next day, Meta: The US district judge Vince Chhabria, in San Francisco, said in his decision on the Meta case that the authors had not presented enough evidence that the technology company's AI would cause "market dilution" by flooding the market with work similar to theirs. Judging by the rulings in favor of Meta and Anthropic, the authors are facing an uphill battle. Three weeks ago, Disney and NBCUniversal sued Midjourney, alleging that the ...

Earlier in the week, Blackburn attempted to forge a compromise with Ted Cruz, who led the provision. Together, they produced a new version that reduced the ten-year ban to a five-year one, and carved out exceptions for laws related to kids' online safety and personal publicity rights. But this version of the bill was promptly excoriated by vocal coalitions in both parties. A group of 140 mostly left-leaning advocacy organizations, including Encode AI and Common Sense Media, penned an open letter arguing that this new version actually shielded tech companies from the state regulation that Blackburn was attempting to protect. "The vague standards set out in the moratorium will provide Big Tech a clear path to challenge nearly any state law in court," the letter read.

A data center for cryptocurrency mining, cloud services, and AI computing in Stutsman County, North Dakota.halbergman/Getty This story was originally published by the Guardian and is reproduced here as part of the Climate Desk collaboration. Republicans are pushing to pass a major spending bill that includes provisions to prevent states from enacting regulations on artificial intelligence. Such untamed growth in AI will take a heavy toll upon the world's dangerously overheating climate, experts have warned. About 1 billion tons of planet-heating carbon dioxide are set to be emitted in the US just from AI over the next decade if no restraints are placed on the industry's enormous electricity consumption, according to estimates by researchers at Harvard University and provided to the Guardian.

On the other side, plaintiffs range from individual artists and authors to large companies like Getty and the New York Times. The outcomes of these cases are set to have an enormous impact on the future of AI. In effect, they will decide whether or not model makers can continue ordering up a free lunch. If not, they will need to start paying for such training data via new kinds of licensing deals--or find new ways to train their models. And that's why last week's wins for the technology companies matter. If you drill into the details, the rulings are less cut-and-dried than they seem at first.

A deal that had been reached between Sens. Marsha Blackburn, R-Tenn., and Ted Cruz, R-Texas, over how states can regulate artificial intelligence has been pulled from President Donald Trump's "big, beautiful" bill. The collapsed agreement would have required states seeking to access hundreds of millions of dollars in AI infrastructure funding in the "big, beautiful" bill to refrain from adopting new regulations on the technology for five years, a compromise down from the original 10 years. It also included carveouts to regulate child sexual abuse material, unauthorized use of a person's likeness and other deceptive practices. Blackburn announced Monday night that she is withdrawing her support for the agreement. A deal between Sens. Marsha Blackburn and Ted Cruz over how states can regulate AI has been pulled from the "big, beautiful" bill.

Tracking devices inserted under offenders' skin, robots assigned to contain prisoners and driverless vehicles used to transport them were among the measures proposed by technology companies to ministers who are gathering ideas to tackle the crisis in the UK justice system. The proposals were made at a meeting of more than two dozen tech companies in London last month, chaired by the justice secretary, Shabana Mahmood, minutes seen by the Guardian show. Amid an acute shortage of prison places and probation officers under severe strain, ministers told the companies they wanted ideas for using wearable technologies, behaviour monitoring and geolocation to create a "prison outside of prison". Those present included representatives of Google, Amazon, Microsoft and Palantir, which works closely with the US military and has contracts with the NHS. IBM and the private prison operator Serco also attended alongside tagging and biometric companies, according to a response to a freedom of information request.

As Congress races to pass President Donald Trump's "Big Beautiful Bill," it's also sprinting to placate the many haters of the bill's "AI moratorium" provision which originally required a 10-year pause on state AI regulations. The provision, which was championed by White House AI czar and venture capitalist David Sacks, has proved remarkably unpopular with a diverse contingent of lawmakers ranging from 40 state attorneys general to the ultra-MAGA Representative Marjorie Taylor Greene. Sunday night, Senator Marsha Blackburn and Senator Ted Cruz announced a new version of the AI moratorium, knocking the pause from a full decade down to five years and adding a variety of carve-outs. But after critics attacked the watered-down version of the bill as a "get-out-of-jail free card" for Big Tech, Blackburn reversed course Monday evening. "While I appreciate Chairman Cruz's efforts to find acceptable language that allows states to protect their citizens from the abuses of AI, the current language is not acceptable to those who need these protections the most," Blackburn said in a statement to WIRED.

This paper envisions a knowledge-guided multi-agent framework named KGMAF for automated requirements development. KGMAF aims to address gaps in current automation systems for SE, which prioritize code development and overlook the complexities of requirements tasks. KGMAF is composed of six specialized agents and an artifact pool to improve efficiency and accuracy. Specifically, KGMAF outlines the functionality, actions, and knowledge of each agent and provides the conceptual design of the artifact pool. Our case study highlights the potential of KGMAF in real-world scenarios. Finally, we outline several research opportunities for implementing and enhancing automated requirements development using multi-agent systems. We believe that KGMAF will play a pivotal role in shaping the future of automated requirements development in the era of LLMs.

Large language models (LLMs) have made significant strides, extending their applications to dialogue systems, automated content creation, and domain-specific advisory tasks. However, as their use grows, concerns have emerged regarding their reliability in simulating complex decision-making behavior, such as risky decision-making, where a single choice can lead to multiple outcomes. This study investigates the ability of LLMs to simulate risky decision-making scenarios. We compare model-generated decisions with actual human responses in a series of lottery-based tasks, using transportation stated preference survey data from participants in Sydney, Dhaka, Hong Kong, and Nanjing. Demographic inputs were provided to two LLMs -- ChatGPT 4o and ChatGPT o1-mini -- which were tasked with predicting individual choices. Risk preferences were analyzed using the Constant Relative Risk Aversion (CRRA) framework. Results show that both models exhibit more risk-averse behavior than human participants, with o1-mini aligning more closely with observed human decisions. Further analysis of multilingual data from Nanjing and Hong Kong indicates that model predictions in Chinese deviate more from actual responses compared to English, suggesting that prompt language may influence simulation performance. These findings highlight both the promise and the current limitations of LLMs in replicating human-like risk behavior, particularly in linguistic and cultural settings.

This work-in-progress investigates the memorization, creativity, and nonsense found in cooking recipes generated from Large Language Models (LLMs). Precisely, we aim (i) to analyze memorization, creativity, and non-sense in LLMs using a small, high-quality set of human judgments and (ii) to evaluate potential approaches to automate such a human annotation in order to scale our study to hundreds of recipes. To achieve (i), we conduct a detailed human annotation on 20 preselected recipes generated by LLM (Mixtral), extracting each recipe's ingredients and step-by-step actions to assess which elements are memorized--i.e., directly traceable to online sources possibly seen during training--and which arise from genuine creative synthesis or outright nonsense. We find that Mixtral consistently reuses ingredients that can be found in online documents, potentially seen during model training, suggesting strong reliance on memorized content. To achieve aim (ii) and scale our analysis beyond small sample sizes and single LLM validation, we design an ``LLM-as-judge'' pipeline that automates recipe generation, nonsense detection, parsing ingredients and recipe steps, and their annotation. For instance, comparing its output against human annotations, the best ingredient extractor and annotator is Llama 3.1+Gemma 2 9B, achieving up to 78% accuracy on ingredient matching. This automated framework enables large-scale quantification of memorization, creativity, and nonsense in generated recipes, providing rigorous evidence of the models' creative capacities.