Brief

New guidance from CSET suggests that small and medium-sized enterprises (SMEs) should focus on achievable first steps for AI adoption due to resource constraints. These steps include assessing AI readiness, establishing personal AI use guidelines, selecting appropriate use cases, setting limitations, and prioritizing protections. The report highlights that employees are already using AI at work without explicit employer guidance, a trend known as 'shadow AI,' which is prevalent in SMEs. AI

A report highlights the severe consequences of AI facial recognition misidentification, detailing cases where individuals lost jobs, homes, and even child custody due to erroneous arrests. The analysis emphasizes that simply improving the accuracy of facial recognition technology is insufficient to address the irreversible societal impacts of these algorithms. It calls for a deeper consideration of how to manage the profound and lasting effects of algorithmic errors on people's lives. AI

Major cloud providers and AI companies are establishing AI certifications to standardize skills and knowledge in the rapidly evolving field. Companies like Microsoft, Google, and AWS, alongside AI leaders such as Anthropic, are offering these certifications. This move aims to provide a clear pathway for professionals to validate their expertise in AI technologies and applications. AI

A new research paper proposes a framework to help semiconductor manufacturing facilities comply with European Union regulations on safety and sustainability. The proposed system uses 'Professional Proxies,' which are role-based agents operating in secure data spaces, to automate compliance reporting. This approach aims to balance corporate data privacy with the need for multi-stakeholder transparency by enabling factories to export verifiable compliance tokens without revealing proprietary manufacturing details. AI

A new paper proposes a framework to help the EU achieve true sovereignty in frontier artificial intelligence development. The paper highlights the EU's current dependence on US and Chinese AI models and supercomputing capacity. It outlines a five-pillar framework to address economic competitiveness, resilience, security, European values, and foreign relations across the AI stack, aiming to guide policy and identify critical gaps. AI

Sam Altman's eye-scanning company, Tools for Humanity, is reportedly undergoing layoffs amidst OpenAI's confidential IPO filing. The company, known for its Worldcoin project which uses iris scans for identity verification and cryptocurrency, has faced significant regulatory and ethical challenges globally. Despite raising $2.5 billion at a $2.5 billion valuation from investors like Andreessen Horowitz, Tools for Humanity is struggling to generate revenue and has been fined in South Korea, with Kenya banning its operations due to privacy and financial concerns. AI

WuXi AppTec has filed a lawsuit against the U.S. Department of Defense in the U.S. District Court for the District of Columbia. The company is challenging its designation as a Chinese Military Company (CMC) and its inclusion on the 1260H list. WuXi AppTec seeks to have this designation declared invalid and to be removed from the list to protect its stakeholders. AI

Tech Wind is now selling IPEVO's Vurbo.AI, a real-time AI translation tool designed to break down language barriers during international meetings and negotiations. Concurrently, the G7 Cybersecurity WG has issued a declaration emphasizing the importance of cybersecurity for SMEs and calling for a transition to post-quantum cryptography. AI

This article outlines the architecture for building generative AI career coaches that comply with the EU AI Act and support remote workers. It details a system using LangChain and serverless Edge functions, incorporating a risk engine and model cards to meet high-risk AI requirements. The proposed solution emphasizes data security and provides a practical, code-driven approach for HR technology stacks. AI

Pony.ai, in partnership with mobility platform Bolt and automaker Stellantis, has launched a Robotaxi testing program in Luxembourg. This initiative aims to evaluate Pony.ai's seventh-generation Robotaxi technology within the local traffic environment, focusing on safety, performance, and regulatory compliance. The project leverages Luxembourg's supportive policies for autonomous vehicle testing. AI

Apple has unveiled an updated Siri with enhanced AI capabilities, aiming to improve user interaction and functionality. In other tech news, Rokid has addressed allegations of its smart glasses being used for illicit recording. Meanwhile, OpenAI is reportedly preparing for an Initial Public Offering (IPO) by submitting confidential filings. AI

The Securities Association of China is launching a call for AI application case studies within the securities industry. These cases will focus on six key areas: operational efficiency, customer experience, decision support, innovation, R&D, and risk compliance, with an additional category for emerging applications like intelligent agents. The initiative aims to identify and promote effective and replicable AI implementations across securities firms. AI

Accessing premium AI subscriptions like Anthropic's Claude Pro and Google's Gemini Pro from Russia is challenging due to payment and country restrictions. While direct purchases are difficult, users can opt for API access through Russian aggregators, offering ruble payments and legal documentation. This API route is presented as a more stable and predictable option for professional use compared to consumer subscriptions. AI

WuXi AppTec, a Chinese biotechnology firm, has filed a lawsuit against the US government. The company alleges that its inclusion on a Pentagon blacklist, which links it to China's military, is arbitrary and unsupported by facts. WuXi AppTec claims this designation has caused significant reputational and operational damage and is seeking its removal from the list. AI

Agentic AI is increasing pressure on fraud detection systems in the UK, with a new initiative called Build East launching in Hong Kong to support agentic AI startups. This development highlights the growing sophistication of AI in both offensive and defensive capacities within the financial sector. AI

Chinese companies are implementing small-scale layoffs as they integrate AI tools to enhance productivity. This approach is being used to avoid government scrutiny and potential social unrest associated with larger workforce reductions. While AI adoption is expected to decrease overall hiring needs, major corporations are proceeding cautiously with gradual reductions and natural attrition rather than direct, large-scale cuts. AI

OpenAI's Apps SDK is evolving beyond simple demo widgets into a framework for building and distributing applications within ChatGPT. This SDK requires a rigorous readiness checklist for internal pilots, focusing on aspects like tool discovery, safe write actions, UI behavior, and administrative controls. For SaaS vendors, the emphasis is on treating these integrations as governed projects rather than marketing launches, ensuring robust server-side logic, clear tool contracts, and secure user authentication before public release. AI

San Diego State University has installed over 1,300 AI-enabled cameras across its campus, including more than 330 in student dormitories, without adequately informing students. The university claims the cameras are primarily for motion detection and that AI features like facial recognition are disabled, despite the hardware's capabilities. This deployment has raised privacy concerns, as the extent and functionality of the surveillance system were only revealed through student journalism and public records requests, contradicting campus policies on privacy in residential areas. AI

GitHub has released a public preview of a REST API for auditing Copilot cloud agent configurations, enabling security reviewers and platform owners to inspect settings before wider adoption. The API allows for detailed checks on repository-level policies, including MCP server inheritance, review tools, workflow approvals, and firewall configurations. This new endpoint is crucial for organizations looking to implement coding agents responsibly by providing a clear governance checklist for readiness. AI

A far-right PAC, Citizens for Sanity, has released an AI-generated attack ad targeting Texas Senate candidate James Talarico. The ad falsely depicts Talarico in a dress, singing a parody of "My Favorite Things," aiming to smear him with fabricated imagery. This tactic is part of a broader strategy by dark-money groups to influence elections through disinformation, with Citizens for Sanity reportedly spending $90 million on similar ads. AI

A Florida man was wrongly imprisoned based on a 93% facial recognition match from a low-quality surveillance image. This incident raises serious questions about the accuracy and reliability of facial recognition technology, particularly in legal contexts. The case highlights the potential for significant miscarriages of justice when such systems are used without sufficient scrutiny. AI

Apple is making a renewed effort to attract developers to its AI initiatives, emphasizing privacy and contextual understanding. Following a challenging period with its AI offerings in 2024 and 2025, the company is now presenting a more polished and capable vision. This strategy aims to bolster developer engagement and ultimately enhance the user experience across Apple's ecosystem. AI

Anthropic is increasingly restricting specific use cases for its Claude models, particularly those involving code generation and agentic capabilities, from its standard subscription tiers. This move appears to be a strategic effort to encourage users to adopt higher-priced plans or enterprise solutions for advanced functionalities. The article suggests this pattern of feature segmentation was anticipated by users and developers who have continued to integrate Claude's code capabilities despite the evolving access policies. AI

The UK's National Health Service (NHS) is expanding its use of Microsoft Copilot, prescribing over half a million licenses to its staff. This decision follows a trial that indicated the AI chatbot could save healthcare workers approximately 43 minutes per day. The expansion aims to alleviate the administrative burden on NHS personnel. AI

An intellectual property expert suggests Nintendo might only receive a $30,000 payout from its legal dispute with Pocketpair, the developer of Palworld. This potential payout is contingent on Nintendo winning the case, though the expert implies the outcome is uncertain. AI

A recent Ninth Circuit ruling addressed the issue of AI-generated hallucinations in legal filings. The court's decision highlights the challenges and potential consequences of relying on AI tools that produce inaccurate information. This case underscores the need for careful verification of AI-generated content, particularly in professional and legal contexts. AI

Thomas Lind, a key figure in the White House's AI policy efforts, is departing his role. Lind served as the head of policy for the Office of the National Cyber Director and was a senior adviser to Director Sean Cairncross. His departure is reportedly for personal reasons, aiming to spend more time with his family. AI

Apple's upcoming iOS 17 will introduce enhanced AI capabilities for Siri, but full support will be limited to newer iPhone models. The company is strategically rolling out these advanced features, likely due to hardware requirements, leaving older devices with a more basic Siri experience. This approach mirrors similar tiered feature rollouts seen with other AI-powered services. AI

Mickai has released eight new briefings detailing AI applications and audit questions across various UK institutions. These cover areas such as NHS coding, the coroner's court, procurement, and local government. The briefings also address the Sovereign AI Fund, bank model risk, and AI safety evaluation, with Mickai signing each action using a post-quantum key. AI

The UK government plans to pilot AI legal assistants in England and Wales' crown courts to help reduce case backlogs. Deputy Prime Minister David Lammy will announce the initiative, which aims to save administrative time and expedite justice. However, legal professionals, including the Law Society, have cautioned that the technology should not be used to cut funding or staff, emphasizing the need for thorough evaluation and robust safeguards against AI hallucinations and fabricated case law. AI

ChatGPT is reportedly set to receive its most significant upgrade to date, with new AI proctors being introduced for college entrance exams to automatically capture unusual video footage. Separately, eight listed companies have announced concentrated tax補繳 (tax補繳) events, indicating a potential shift in tax compliance or audits. AI

A far-right political action committee has spent over six figures to create an AI-generated image of Texas Senate candidate James Talarico in a dress. The fabricated image, which lacks any disclaimer, was set to a song titled "My Favorite Things." This tactic is being used as opponents cannot defeat Talarico on policy, as he currently leads indicted AG Ken Paxton by three points. AI

The State Commissioner for Data Protection in Baden-Württemberg has released two practical guides aimed at local government administrations. These guides address the integration of data protection principles with digitalization efforts, specifically focusing on the use of AI for transcribing municipal council meetings and the application of Section 3a of the State Data Protection Act (LDSG) for AI deployment within public administration. AI

xAI has been accused of firing an engineer who voiced concerns about the safety of its Grok chatbot. The lawsuit claims the engineer was terminated after raising internal alarms regarding the AI's potential risks. This action has led to legal proceedings against Elon Musk's AI company. AI

Anthropic has restricted access to its Claude models for users engaged in life science research. This move aims to prevent the misuse of AI in sensitive areas like drug discovery and biological research. The company is implementing stricter usage policies to ensure responsible AI deployment. AI

Google and YouTube are facing a lawsuit over allegations that they used copyrighted music to train AI models without proper authorization. The lawsuit, filed by musicians and music rights holders, claims that the companies violated copyright laws by scraping and utilizing musical content for AI development. This legal challenge highlights the growing concerns surrounding the use of creative works in AI training and the need for clearer regulations in the digital age. AI

Fintech firm Pagaya has filed a lawsuit against its former partner, Klarna, alleging the company stole its AI-powered subprime lending model. Pagaya claims Klarna misappropriated trade secrets and violated licensing and loan sale agreements. This legal action stems from their previous partnership in the buy-now-pay-later sector. AI

OWASP has released its "Top 10 for Agentic Applications 2026," a new security taxonomy specifically for AI agents. This initiative aims to address the unique security challenges posed by AI agents, such as the potential misuse of application passwords. One of the identified risks, ASI03 "Identity & Privilege Abuse," highlights concerns about how AI agents might exploit existing security measures. AI

AI red teaming, a practice focused on identifying and mitigating risks in AI systems, has significantly matured since its inception. Initially a niche discipline, it has evolved to become a critical component of AI development and deployment. This evolution is driven by the increasing complexity and widespread adoption of AI technologies, necessitating robust safety and security measures. AI

A surveillance company named Leonardo is developing a new technology called SignalTrace that will enhance its automatic license plate readers (ALPRs). This upgrade will allow the ALPRs to not only capture license plates but also detect and record unique identifiers from Bluetooth-enabled devices like phones, AirPods, and smartwatches within passing vehicles. The collected data will be used to create an "electronic fingerprint" to help law enforcement track individuals, even if they change license plates. AI

A new tool called mcp-repo-graph has been introduced to the MCP marketplace, offering developers a way to visualize codebase structure and dependencies. This tool runs locally and does not require external authentication, making it a low-risk option for code navigation. The article also highlights the popularity of official integrations like GitHub Copilot MCP and OpenAI MCP, emphasizing the need for robust allowlisting strategies that account for policy drift across different Integrated Development Environments (IDEs) and potential token cost increases. AI

The UK government has launched a new AI-powered tool called the Work Assistant to help jobseekers craft their applications. This initiative aims to address fears about AI displacing human workers by providing a resource for individuals to create machine-written applications. The tool is intended to assist jobseekers in applying for positions around the clock. AI

A man is suing Florida law enforcement after being arrested based on a facial recognition match that was only 93% accurate. He alleges that the police disregarded other evidence and relied solely on the flawed AI identification. This case highlights concerns about the accuracy and potential misuse of facial recognition technology in law enforcement. AI

An attorney involved in a Mississippi legal case faced disciplinary action for AI-related errors, including a claim of ignorance regarding AI hallucinations. This marks a repeat offense, as the attorney had been previously disciplined for similar issues. The penalty was deemed insufficient by some observers, who suggested a significantly larger fine would have been more appropriate. AI

Microsoft has restricted the use of Anthropic's new Claude Fable AI model for its employees due to data retention concerns. The model, described as Anthropic's first 'Mythos-class' AI, was recently released. This internal restriction highlights ongoing worries about how AI models handle and retain sensitive data within large organizations. AI

Anthropic's new Fable 5 model for enterprise users now retains prompts for 30 days for safety monitoring. Users can opt out of this by using older Claude models that offer zero prompt retention. The effectiveness of these safety measures and how users adapt to them will be a key indicator for their utility in sensitive fields like security and biology research. AI

Flathub, a popular Linux application store, has updated its policy to ban new applications generated by AI. This ban extends to automated submissions, including code, metadata, and build scripts. However, existing AI-generated applications and mature projects will be exempt, with exceptions determined by maintainer review based on factors like community engagement and release activity. AI

AWS has introduced Cross-Region Inference (CRIS) on Amazon Bedrock to enhance AI flexibility in Europe. This feature allows generative AI applications to route requests across multiple AWS Regions within defined geographic boundaries, optimizing model throughput and latency. CRIS helps customers meet data processing and privacy requirements, such as GDPR, by enabling model access while respecting regional intricacies and ensuring application resilience. AI

NHS England is set to deploy Microsoft Copilot to over half a million clinicians and support staff, following a successful pilot program. The AI assistant reportedly saved participants an average of 43 minutes per day on administrative tasks. This large-scale rollout aims to alleviate the significant paperwork burden faced by healthcare professionals. AI

A new rubric for Explainable AI (XAI) in autonomous driving safety has been proposed, highlighting a significant gap between current XAI methods and the evidence required by safety standards. The proposed rubric, derived from automotive safety standards like ISO 26262, identifies that causal XAI methods are structurally necessary for critical stages such as hazard identification and incident investigation. The research suggests that XAI method selection should prioritize the evidence demands of specific lifecycle stages rather than relying on method popularity. AI