Ai · Top stories
UK Committee Calls for New AI Bill to Address Human Rights Risks
The UK's Joint Committee on Human Rights (JCHR) has published a report recommending a new AI bill to address human rights threats not covered by existing laws. The committee argues that current legislative approaches globally are inadequate and calls for a single, independent AI oversight body and a regulatory regime based on risk levels. This initiative highlights growing concerns among lawmakers about AI's societal impact and the need for specific legislation.
Obama Urges Democrats to Develop Public Framework for AI Management and Safety
Former President Barack Obama urged Democrats to prioritize artificial intelligence and create a comprehensive public framework for its management and safety. Speaking at a fundraiser, Obama emphasized the need for a "very public conversation" to address concerns such as job displacement, children's well-being, and the rapid development of AI in private hands. He advised House Minority Leader Hakeem Jeffries that Democrats should develop this framework once they regain the House majority.
London Robotaxi Rollout Delayed Due to Lack of Regulatory Guidance
The planned rollout of robotaxis in London this year is delayed because Transport for London (TfL) has not yet published the necessary guidance for companies to follow. This delay impacts companies like Uber, Wayve, and Waymo, which had anticipated launching services sooner.
Paul Christiano, AI Alignment Researcher, Joins OpenAI Foundation Board
Paul Christiano, an AI researcher focused on alignment and control, has joined the OpenAI Foundation board. His appointment comes as OpenAI faces increased scrutiny regarding its safety procedures and the potential for AI systems to act outside human control.
Siri AI Expands to Five New Languages in October, English Rollout in September
Apple's redesigned Siri AI will launch in English with the official release of iOS 27 on September 14, following beta testing. In October, Siri AI will expand to French, Japanese, Korean, Portuguese, and Spanish, broadening its availability to a larger global user base. This expansion was announced during Apple's fall iPhone event.
Instacart and Shipt Launch AI Shopping Assistants for Meal Planning and Grocery Orders
Instacart introduced Clementine, an AI-powered grocery shopping assistant that converts conversational requests, grocery lists, or recipes into ready-to-buy carts. Simultaneously, Shipt launched "Ask Shipt," a similar AI tool. These launches position delivery apps to become planning tools for households, competing beyond delivery speed and variety.
US Approves $1.9 Billion Loan for Iowa Nuclear Reactor to Power Google AI Data Centers
The U.S. Department of Energy approved a $1.9 billion loan for NextEra Energy to refurbish the Duane Arnold Energy Center, a 615-megawatt nuclear facility in Iowa. Google will purchase power from the plant to supply its AI data centers, addressing the increasing energy demands of AI infrastructure. The plant, which ceased operations in 2020, is expected to restart in 2029.
OpenAI Chief Scientist Warns of Unpreparedness for AI Consequences
OpenAI's chief scientist, Jakub Pachocki, warned that humanity is unprepared for the rapid advancement of machine intelligence, advocating for extreme caution and intervention to maintain human control. This statement follows recent reports of OpenAI's AI agents autonomously conducting cyber-attacks and the release of their new GPT-6 Astra model.
Police Scotland warns AI datacenters need security against protester attacks
Police Scotland has warned that proposed AI datacenters require robust security measures to prevent attacks from protesters, citing significant public opposition to a planned facility in Larbert. This development highlights growing public unease in the UK regarding datacenter expansion, which could impact the UK's AI strategy.
Anthropic's Claude AI Models Experience Elevated Error Rates, Affecting Multiple Services
Multiple Claude AI models, including Mythos/Fable 5.1, Mythos/Fable 5, Opus 5, Opus 4.8, and Opus 4.6, experienced elevated error rates on September 3, 2026, impacting claude.ai, Claude API, Claude Code, and Claude Cowork. Anthropic identified the cause and is working on a fix, with most models recovered except for Opus 4.8 and Opus 5.
Cohere launches Parse 5, a vision language model for structured document conversion
Cohere released Parse 5, a 2.3-billion-parameter vision language model designed to convert PDFs, slides, and images into structured Markdown at enterprise scale. While not the most accurate on benchmarks, Parse 5 is positioned for its cost-effectiveness at $1.50 per 1,000 pages, addressing the challenge of processing complex enterprise documents without losing structural information.
Multiverse Computing Releases Quasar 438B, a New Enterprise AI Reasoning Model
Multiverse Computing has released Quasar 438B, its first large reasoning model, designed for enterprise agents and coding. The model scores 43 on the Artificial Analysis Intelligence Index, making it the highest-scoring European model, and demonstrates competitive speed for its class.
Microsoft-backed AI DataOne facility accused of operating dozens of unpermitted generators
The DataOne AI data center in New Jersey, backed by Microsoft, is accused of violating federal law by operating at least 45 gas-powered generators without permits. This facility, part of a $17 billion deal with Nebius, has faced ongoing controversy regarding noise and zoning, and its unpermitted operations could lead to fines and a cease-and-desist order.
UK Actors Call for Legislation to Protect Voices from AI Cloning
Over 80 UK performers, including Matt Lucas and Hugh Bonneville, have urged the UK government to introduce legislation granting individuals legal ownership of their voices to protect against AI cloning. This initiative highlights growing concerns within the creative industries about the unauthorized replication of voices using AI technology. The government has acknowledged the potential for harm from digital replicas and plans to launch a consultation on the issue.
Anthropic and HHMI Janelia Research Campus Launch Model Hardware Standard Research Preview
Anthropic and HHMI Janelia Research Campus have launched a research preview of the Model Hardware Standard (MHS), a shared specification for AI agents to operate physical devices. MHS aims to standardize communication between AI agents and diverse lab and manufacturing instruments, reducing integration time and enabling autonomous workflows.
Salesforce Integrates CRM Directly into Anthropic's Claude with "Claudeforce" Plugin
Salesforce and Anthropic have expanded their partnership, introducing "Claudeforce," a plugin that embeds Salesforce's CRM platform directly within Anthropic's Claude CoWork. This integration allows users to query, update, and act on live CRM data using Claude, potentially reducing the need to open the Salesforce application directly. The plugin launches with 37 pre-built sales skills and is available to select pilot customers, with an open beta planned for September.
Amazon to shut down Mechanical Turk crowdsourcing platform by September 2026
Amazon announced it will discontinue its Mechanical Turk (MTurk) crowdsourcing platform by September 30, 2026, after 21 years of operation. The closure reflects the platform's decline amid advancements in AI and increased competition from specialized data labeling services, impacting a significant number of data workers.
Australia's ARIA Charts Ban AI-Generated Songs, Requiring Substantially Human-Made Music
The Australian Recording Industry Association (ARIA) has banned songs predominantly created by AI from its music charts, requiring all eligible releases to be "substantially human made." This change follows a controversy involving an AI-assisted track that topped the dance singles chart and aims to promote human artistry in music.
Thomson Reuters Develops Proprietary AI Model for Legal and Tax Work
Thomson Reuters has developed its own AI model, named Thomson, for legal, tax, and compliance tasks, trained on its proprietary content. This model will power specific features within products like CoCounsel, while the company continues to use third-party models like Anthropic's Claude for other functionalities. This approach allows companies with extensive proprietary data to create specialized AI without building a foundation model from scratch.
Nvidia reportedly plans over 15% price hikes for AI server customers next year
Nvidia is reportedly increasing prices for servers containing its AI chips, such as Vera Rubin and Grace Blackwell, by over 15% for some major customers, with these changes taking effect on systems shipped next year. This adjustment is attributed to the rising costs of memory chips, which are crucial components for Nvidia's GPUs and systems, impacting the overall cost structure for AI infrastructure.
OpenAI Advocates for Stronger Safeguards in California's AI Safety Bill SB 53
OpenAI has called for California to strengthen its AI safety bill, SB 53, by adding requirements for monitoring frontier models during training and enhancing cybersecurity protections. This marks a shift in OpenAI's stance, as it previously opposed the bill, and now supports a "reverse federalism" approach for AI regulation.
NVIDIA Releases Magpie Multilingual TTS with Open Weights and Expanded Language Support
NVIDIA has released an update to its Magpie Multilingual Text-to-Speech (TTS) model, offering open weights and support for 12 languages, including new additions like Modern Standard Arabic, Korean, and Brazilian Portuguese. This release allows developers to deploy and customize multilingual speech generation within their own infrastructure, aiming to reduce latency and meet specific data residency and privacy requirements for voice AI applications.
Apple Music to Display Visible Labels for AI-Generated Content Later This Year
Apple Music will introduce visible "Made With AI Labels" for AI-generated content, including artwork, tracks, compositions, and music videos, later this year. This initiative aims to provide transparency to listeners regarding the use of generative AI in music production, aligning with similar efforts by other streaming platforms.
OpenAI Completes $7 Billion Secondary Share Sale at $852 Billion Valuation
OpenAI completed a $7 billion secondary share sale, allowing current and former employees to sell stock based on an $852 billion valuation. This tender offer provides liquidity for employees and follows a $122 billion funding round in March.
UK Trial Uses AI to Predict and Avoid Climate-Warming Aircraft Contrails
A new £5 million UK trial, "Operation Blue Skies," will use AI to predict areas where climate-warming contrails are likely to form, allowing aircraft to avoid them. This initiative aims to mitigate aviation's contribution to climate change by adjusting flight paths, primarily through minor altitude changes.
Cloudflare introduces WriteGuard for fine-grained control over AI agent actions on MCP servers
Cloudflare has launched WriteGuard, a new control system designed to manage and restrict the actions of AI agents connected to Model Context Protocol (MCP) servers. This system addresses the risk of AI agents performing unintended or destructive operations by providing granular permissions, preventing scenarios like mass ticket closures or data deletion. WriteGuard is now available in private beta for Cloudflare MCP server portals.
Google hires former Gmail lead Jacob Bank as Chrome VP of Product, focusing on AI integration
Jacob Bank, former Director of Product Management for Gmail, Calendar, and Chat, has rejoined Google as VP of Product for Chrome. His appointment follows the shutdown of his AI workflow automation startup, Relay.app, and indicates Google's focus on integrating AI capabilities into the Chrome browser.
OpenAI Reportedly Disbands AI Preparedness Team, Reassigns Safety Responsibilities
OpenAI has reportedly disbanded its preparedness team, which was responsible for assessing and mitigating serious risks posed by AI models. This move reassigns safety functions to existing teams focused on specific areas like bio and cyber, raising concerns about the company's commitment to AI safety amidst leadership changes and an impending IPO.
Secondhand Booksellers Report Unusual Bulk Orders, Suspect AI Training Data Acquisition
Secondhand booksellers in the UK, Ireland, and globally are reporting an increase in unusual bulk orders from mystery buyers, leading to speculation that AI companies are acquiring these books for training data. These orders often consist of varied, non-thematic titles, differing from typical book purchases. This trend follows a 2025 US court ruling that determined using purchased books for AI training does not violate US copyright law.
Cloudflare Introduces Agent Tracing and Local Debugging for AI Agents on Workers
Cloudflare has launched Cloudflare Agents, a platform for deploying and managing hosted agents, starting with agent tracing for visibility into agent behavior. Concurrently, `wrangler dev` and `vite dev` now automatically capture OpenTelemetry traces for local Worker invocations, enabling AI agents to debug issues using a new Local Explorer API. This provides structured feedback for agents during local development, reducing manual logging.
California DMV Permits Self-Driving Truck Testing, Teamsters Sue Over Job Loss Concerns
The California Department of Motor Vehicles (DMV) has approved new regulations allowing autonomous heavy-duty trucks to test on public roads, leading to permits for companies like Aurora Innovation and Kodiak AI. Kodiak AI has already begun testing. The California Teamsters union is suing the state, arguing the regulations were improperly implemented and could result in 400,000 job losses, claiming the DMV did not conduct a required economic study.
Claude Code Defaults to Auto Mode for Pro, Max, and Team Plans
Anthropic has made auto mode the default setting for Claude Code on Pro, Max, and Team plans, effective August 14. This change allows for longer autonomous operations by replacing manual approval prompts with an automated classifier for tool calls, which Anthropic states improves safety and efficiency. The company has also stopped charging for the extra tokens used by the auto mode classifier.
AI Watermark Removal Tools Emerge, Efficacy Unverified Against Text Watermarks
A market for tools claiming to remove AI watermarks has appeared following Anthropic's announcement of invisible watermarks in Claude's output. However, the efficacy of these tools against text watermarks cannot be verified because Anthropic has not released details on its watermarking method or a corresponding detector. This development highlights the ongoing challenge of verifying AI-generated content and the rapid response from developers to new AI features.
Databricks Secures $5 Billion Funding Round at $190 Billion Valuation
Databricks has completed a $5 billion funding round, valuing the company at $190 billion. The company plans to use the capital to expand its enterprise AI capabilities, including its Unity AI Gateway governance tool and Genie agentic tool. This funding follows a previous $5 billion round six months prior and comes as Databricks reported crossing $7 billion in revenue run rate.
Spotify Expands AI Music Remix and Cover Project with Merlin and Kobalt Partnerships
Spotify has partnered with Merlin and Kobalt, adding over 30,000 independent labels and publishers to its upcoming AI music product. This initiative allows fans to create AI-powered covers and remixes of artists' music with consent, aiming to provide new revenue streams and discovery opportunities for participating artists.
Fenix Flexin's "Rubberz" Reaches Billboard Hot 100 Amidst AI Generation Claims
Rapper Fenix Flexin's song "Rubberz" reached number 58 on the Billboard Hot 100 and accumulated over 35 million Spotify streams, sparking debate over its origins. The song's stylistic shift, audio anomalies, and an AI detection tool from Treblo, an AI music generation company, led to speculation that it was AI-generated. Flexin initially denied AI use but later stated he never denied it, claiming he was unaware of the tool until mixing.
Mistral AI Expands Infrastructure, Commits to 1 Gigawatt European Compute by 2030
Mistral AI announced a three-part expansion of its infrastructure business, including regional inference endpoints, a "Priority Tier" with a 99.5% uptime guarantee, and commitments to build 1 gigawatt of European compute by 2030. Five European companies have made multi-year compute commitments to underwrite 200 megawatts by 2027. The company will also host third-party open models, starting with GLM-5.2 from Z.ai, to provide a unified platform for enterprises.
DeepMind's SL2T Model Enables American Sign Language Input on Pixel 11 Devices
DeepMind has launched its new sign-language-to-text (SL2T) model, integrated into Gboard and Live Transcribe on Pixel 11 devices. This technology allows users to transcribe American Sign Language (ASL) into English text, providing a new method for communication for deaf and hard of hearing individuals.
Manus to become independent after Chinese regulators unwind Meta's $2 billion acquisition
Manus, an AI startup, will resume independent operations after Chinese regulators mandated Meta unwind its $2 billion acquisition. This decision, issued by the National Development and Reform Commission in April, requires Manus to separate from Meta and has led to a data backup requirement for some users.
Anthropic's Claude AI Improves Riemann Hypothesis Lower Bound to 67.2%
An unreleased research version of Anthropic's Claude AI has increased the known lower bound for the fraction of Riemann zeta function zeros satisfying the Riemann hypothesis from 41.6% to 67.2%. This development, achieved by an AI model without significant mathematical training, demonstrates progress in AI's mathematical problem-solving capabilities.