California State Senator Scott Wiener has introduced a new bill requiring companies developing artificial intelligence (AI) models with high computing performance thresholds to publicly disclose their safety and security protocols, aiming to mitigate potential catastrophic risks posed by AI technology.
https://news.bloomberglaw.com/esg/california-lawmaker-pushes-ai-firms-to-release-safety-policiesResearchers have raised concerns about ChatGPT's reliance on potentially manipulated sources, highlighting the need for improved detection and filtering of deceptive content to maintain trust in AI-generated recommendations. This issue underscores the importance of building authority through trustworthy content and earned media as platforms like ChatGPT become increasingly influential in user discovery.
https://www.searchenginejournal.com/chatgpt-recommendations-potentially-influenced-by-hacked-sites/550788/Chartered Professional Accountants (CPAs) are stepping in to address growing concerns around artificial intelligence (AI) by providing trusted oversight through their assurance services. A joint publication from CPA Canada and the American Institute of CPAs highlights the need for independent validation of AI systems, citing risks such as financial loss, regulatory penalties, and reputational damage. According to Melissa Robertson, a CPA Canada expert, this is particularly pressing in Canada, where AI literacy ranks low. By applying established processes and professional standards, CPAs aim to build trust through assurance services that offer third-party validation of complex systems' design and operation.
https://www.manilatimes.net/2025/07/09/tmt-newswire/globenewswire/accountants-play-a-critical-role-in-building-trust-in-ai/2146098Logistics firms in Singapore are leveraging artificial intelligence solutions like chatbots and autonomous robots to enhance efficiency amid rising e-commerce volumes. FedEx has partnered with QuikBot Technologies to deploy the Quikbot, an autonomous final-mile delivery platform that automates tasks such as navigating elevators and crossing gantries, allowing couriers to focus on customer interaction. The system uses digital mapping of buildings to program the robots' routes, reducing waiting time for couriers like those at South Beach Tower, where trials have saved over an hour of work per day.
https://channelnewsasia.com/singapore/autonomous-robots-ai-chatbots-logistics-delivery-ecommerce-5228266Gmail's new AI-powered feature allows users to suggest responses to emails before hitting send, but experts warn that relying too heavily on AI-generated content can lead to awkward or unprofessional responses, as seen in a recent example where an AI-suggested response was misinterpreted by the recipient. Google has since updated its guidelines to advise users to review and edit AI-generated content before sending it.
https://nypost.com/2025/07/09/lifestyle/people-are-fuming-over-this-lazy-job-rejection-email/Google's business-as-usual approach with Israel has been criticized for contributing to a "joint criminal enterprise" that enables genocide. The company's work on Project Nimbus, a billion-dollar cloud computing and AI project for the Israeli government and military, has sparked controversy among employees. In response to Hamas' October 7, 2023 attack, Google rushed to sell more AI tools to Israel. This decision led to protests and sit-ins by some employees, resulting in about 50 firings. Google dropped a prior pledge not to use AI in surveillance tools or weapons, potentially opening new profit avenues at the expense of those targeted.
https://gizmodo.com/googles-sergey-brin-calls-u-n-report-on-big-techs-relationship-with-israel-antisemitic-2000626823The shift from traditional data extraction and aggregation (RAG) to software agents that query source systems directly at runtime offers several advantages, including eliminating duplicate data repositories, preserving authorization models, and reducing the attack surface. However, enterprise implementations of agent-based architectures are more measured, with specific workflows, explicit permission models, and comprehensive audit trails. To ensure security, organizations must implement robust authentication and authorization controls, visibility and monitoring systems, and content protection mechanisms. A case study by Grand Canyon Education demonstrates the effectiveness of API-driven security guardrails in securing AI chatbot platforms, highlighting the importance of adapting approaches to address security, performance, and user experience challenges.
https://www.techradar.com/pro/rag-is-dead-why-enterprises-are-shifting-to-agent-based-ai-architecturesIBM's Power11 servers boast a remarkable 99.9999% uptime, ensuring minimal downtime throughout their lifecycles. The new hardware is designed to support AI, hybrid cloud, and automation applications with enhanced performance and security features, including autonomous patching, live updates, and rolling upgrades. Additionally, the Power11 servers incorporate quantum-safe cryptography to prepare against future cyberattacks, guaranteeing less than one-minute detection of ransomware threats.
https://www.techradar.com/pro/ibm-unveils-first-new-power-servers-for-years-to-boost-ai-and-securityiMerit has launched a global network of expert scholars to provide secure and high-quality model tuning for Generative AI foundation models and applied AI customers. The platform, which focuses on expertise in infrastructure, curates thousands of cognitive specialists with advanced degrees in computer science, medicine, biology, finance, law, and policy. These experts can handle multilingual or multi-domain capabilities, providing critical human oversight to differentiate AI systems through specialized knowledge and expert prompting and reasoning inputs.
https://www.manilatimes.net/2025/07/09/tmt-newswire/pr-newswire/imerit-unveils-scholars-a-handpicked-global-network-of-cognitive-experts-for-advanced-genai-training/2146486