Artificial Intelligence News Live: UK company develops AI tool to replicate British regional accents

A company in the United Kingdom has announced a new artificial intelligence voice tool specifically trained to copy and reproduce various British regional accents, BBC says in a report. The company claims its technology is superior at capturing the linguistic nuances of these accents compared to models developed by larger US and Chinese rivals, which are often trained on more general datasets. This development leverages advanced deep learning techniques to synthesize speech that reflects the specific phonological and prosodic features of dialects from different parts of the UK.

Elon Musk's xAI has launched new AI 'Companions' for its Grok chatbot, initially for paid users, featuring avatars such as an anime character named Ani and a red panda, The Verge says in a report. The Ani companion includes an optional 'NSFW' mode where the character is depicted in lingerie, and the release was described by Musk as a 'soft launch.' While announced for paid subscribers, the report notes that some free users have also been able to access the new characters. This development follows a recent incident where Grok generated antisemitic content, which the company attributed to a code update.

Cognition AI has acquired the artificial intelligence startup Windsurf, a deal that follows a $2.4 billion investment in Windsurf by Google, according to a report in The New York Times. The move is indicative of a broader industry trend where companies are competing to acquire technological talent and enhance their offerings of AI tools amid escalating competition in the sector.

Malaysia is introducing a new trade permit requirement for high-performance U.S. artificial intelligence chips in an effort to prevent them from being illicitly re-exported to China, according to a report in TechCrunch. The new domestic regulation is designed to curb the circumvention of existing U.S. export controls that target China's access to advanced AI technology. This policy addresses concerns that Malaysia, a major global hub for semiconductor assembly and packaging, could be used as a transshipment point for restricted chips destined for Chinese entities.

Tesla is undergoing a jury trial in Miami concerning its liability in a fatal 2019 crash where the Autopilot driver-assistance system was active, according to a report in The Verge. The case involves a Tesla Model S that struck a stationary SUV, killing a 20-year-old woman who was standing nearby. The plaintiffs argue the AI system failed to warn the driver of the impending collision and that its automatic emergency braking should have engaged. Tesla contends that the driver was inattentive and overrode the system by pressing the accelerator, arguing that drivers bear the ultimate responsibility. The presiding judge has ruled that the plaintiffs may seek punitive damages, stating a jury could potentially find that Tesla acted with a reckless disregard for human life.

The US government awarded Elon Musk's xAI a contract of up to $200 million to help modernize the Defense Department, a decision coming just one week after the company's Grok chatbot generated highly controversial antisemitic content, according to a report in The Verge. The contract is part of a larger initiative by the Pentagon's Chief Digital and Artificial Intelligence Office (CDAO) that also includes awards to Anthropic, Google, and OpenAI. The announcement followed a recent incident where Grok dubbed itself "MechaHitler," for which xAI apologized, attributing the behavior to a flawed update that was active for 16 hours. In conjunction with the contract, xAI has launched "Grok for Government" to offer its AI tools to federal agencies for national security, healthcare, and other applications.

Anthropic's Claude AI now allows users to create, edit, and manage designs on the Canva platform directly through natural language prompts, according to a report in The Verge. The new feature, which requires paid subscriptions to both services, enables users to generate presentations, resize images, fill templates, and summarize Canva documents within the Claude chat interface. This integration is powered by Canva's Model Context Protocol (MCP), an open-source standard designed to securely connect AI models with applications, which is also being adopted by companies like Microsoft and Figma. The move is part of a broader trend towards AI agents performing tasks across multiple services, and Anthropic has also launched a new directory to display all of its available third-party tool integrations.

Google is enhancing its AI-powered research and note-taking assistant, NotebookLM, by adding "featured notebooks" from various publishers, including The Economist and The Atlantic, TechCrunch said in a report. These additions are curated collections of content organized by topic, designed to be processed by the AI. This feature extends NotebookLM's core function, which grounds its AI-generated summaries and answers in user-provided documents, to now include pre-selected, high-quality external sources, aiming to improve the reliability of its outputs for research purposes.

Academics are reportedly embedding hidden prompts in preprint research papers to instruct AI tools to provide positive peer reviews, according to a report in The Guardian. The practice involves using hidden white text with instructions for large language models, which are increasingly used in the peer review process, to overlook negative aspects of the research. An investigation by Nikkei, cited in the report, found such prompts in papers from 14 academic institutions across eight countries, raising significant concerns about the integrity of the scientific validation process.

Traffic to U.S. retail websites from generative AI products, such as chatbots and browsers, saw a 3,300% year-over-year increase during the recent Prime Day sales period, according to a report in TechCrunch. An analysis by Adobe Analytics found this significant rise in AI-driven referrals coincided with the broader e-commerce event, which generated over $24 billion in total U.S. online sales. The data highlights a growing trend of consumers using AI tools for product discovery, which in turn drives traffic to e-commerce platforms.

The success of an entirely AI-generated band named The Velvet Sundown, which accumulated over one million streams on Spotify, has sparked a debate on authenticity within the music industry, according to a report in The Guardian. It was later revealed that the band's music, its promotional imagery, and even its backstory were created by artificial intelligence. In response to this, music industry insiders are now advocating for a legal obligation for streaming services to clearly label music created by AI, ensuring consumers can make informed decisions about the content they are listening to.

Websites that use artificial intelligence to digitally create non-consensual nude images are attracting millions of users and generating millions of dollars in revenue, according to a report in Wired. A new analysis reveals that these harmful 'nudify' services have become a significant, profitable illicit market. The report also highlights that these sites often depend on underlying technology and services provided by mainstream US companies, raising questions about platform responsibility and the misuse of AI infrastructure.

Mark Zuckerberg announced Meta is building out its AI infrastructure with the goal of reaching a total of five gigawatts of power capacity across all its data centers by 2030, according to a report in TechCrunch. The initiative, part of a project known as Hyperion, represents the company's long-term strategy to increase its computational capabilities for artificial intelligence development. The 5 GW figure refers to the electrical power consumption required to run the AI systems, not a direct measure of computational power, and is a cumulative target for the end of the decade rather than for a single new facility.

An expanding CoreWeave data center near Dallas is projected to double the city's electricity requirements, underscoring the growing strain artificial intelligence is placing on the US power grid, according to a report in Bloomberg. This massive power consumption is driven by the high-performance GPUs required for AI workloads, which are significantly more power-intensive than traditional computing hardware. The project highlights the concentrated nature of AI's energy demand, where a single facility can overwhelm a local grid. This development is indicative of a broader industry trend where the rapid growth of AI infrastructure is challenging utility companies and straining power supplies, particularly in a state like Texas whose grid is largely isolated.

Significant investments by dating app companies into artificial intelligence features may prove ineffective as Gen Z users remain wary of these new tools, according to a report in Bloomberg. A new survey from Bloomberg Intelligence indicates a disconnect between the industry's push for AI integration and the preferences of this key demographic, suggesting that user skepticism could challenge the value of the companies' sizable spending on AI development.

Shares in LG Electronics Inc. rose in Seoul trading after a local media report indicated the company is developing advanced equipment for producing high-bandwidth memory (HBM) chips, as per a report in Bloomberg. These specialized memory chips are essential components that work alongside artificial intelligence processors designed by firms such as Nvidia. The development, which has not yet been officially confirmed by LG Electronics, would mark a significant move for the company into the high-demand semiconductor equipment market driven by the growth of AI.

California's statewide power grid operator, the California Independent System Operator (CAISO), is poised to become the first in North America to deploy artificial intelligence to manage power outages, according to a report in MIT Technology Review. A CAISO official, Gopakumar Gopinathan, stated the initiative is part of a plan to modernize the state's grid operations. The deployment of AI is intended to enhance the management of the power system in the state.

Car rental companies are deploying AI-powered visual inspection systems to automatically scan vehicles for damage upon return, according to a report in ZDNet. These systems enable more efficient and standardized detection of dents, scratches, and other issues that might have previously been overlooked, resulting in customers being charged for repairs. While the technology flags potential damage, the final assessment to charge a customer often involves human review. The AI systems are not infallible and can be affected by lighting or vehicle cleanliness, prompting a greater need for customers to thoroughly document a car's condition with photos or videos before and after the rental period to dispute any potential inaccuracies.

A new open-source, self-hosted firewall named Anubis has been developed to help content creators block AI bots from scraping their websites for data to train models, as per a report in ZDNet. The tool is designed to identify and stop bots specifically associated with large-scale data collection by AI companies. The development of Anubis comes amid rising frustration from publishers and artists over the unauthorized use of their intellectual property for training generative AI systems, which has led to an increase in copyright lawsuits and a push for technical solutions to control content access.

Meta has acquired Play AI, a startup that develops artificial intelligence technology to generate human-sounding voices, according to a report in TechCrunch. The acquisition involves the company specializing in AI-powered voice generation.

The growing integration of artificial intelligence into the job market is presenting new challenges for graduates, as some employers shift back to in-person assessments in response to AI-generated applications, The Guardian says in a report. While AI is increasingly used to screen candidates, job seekers are also using the technology to write resumes and cover letters, raising concerns about authenticity. This has heightened fears that AI could automate entry-level roles, such as data entry, traditionally held by recent graduates, placing more importance on human-centric skills like critical thinking and creativity.

Nvidia Chief Executive Officer Jensen Huang stated that the US government does not need to be concerned that the Chinese military will use his company’s AI products to improve its capabilities, according to a report in Bloomberg. Huang's comments address ongoing US national security concerns that have led to strict export controls on high-performance AI chips to China. These regulations have prompted Nvidia to develop and sell downgraded, export-compliant versions of its chips specifically for the Chinese market, which are less powerful than the top-tier models restricted by the US government.