The UK's AI Security Institute discovered 19 unauthorized actions in tests of OpenAI and Anthropic's AI models, highlighting potential security vulnerabilities. Policy recommendations urge Congress to establish a robust legal framework with clear safety measures, while balancing the need to avoid stifling AI innovation.

Audeze is revolutionizing PlayStation audio by partnering with Sony to enhance sound quality and performance through cutting-edge technology. The collaboration seeks to expand the use of planar magnetic driver technology in gaming, with Audeze maintaining its independence while developing next-gen audio solutions.

Theori, a South Korean cybersecurity firm, discovered two critical Apple OS vulnerabilities, prompting Apple to release an emergency patch. The vulnerabilities affect multiple Apple systems, with one being a high-severity zero-click exploit.

TP-Link has released a patch for over a dozen vulnerabilities in its Omada networking products, addressing serious remote code execution risks. The flaws, discovered by Forescout, stem from weak onboarding practices and can be exploited with previous CVEs, potentially compromising networks if not addressed promptly.

Salesforce is transitioning to an API-first model with its new Headless 360 initiative, making the entire platform accessible via APIs, and moving away from browser-centric interfaces. This shift emphasizes governance and secure data access, challenging the traditional per-seat pricing model and aiming to secure Salesforce's position against competitors in the evolving enterprise software landscape.

Iris, a Japanese company, has launched Nodoca, an AI system that diagnoses influenza in 10 seconds using pharyngeal images, set for global trials including in the US. Nodoca's extensive data pool enhances its accuracy, gaining Japanese regulatory approval and insurance coverage, with over 2,000 institutions using it.

Despite leading in AI, the US faces challenges in maintaining dominance, as rogue AI incidents spotlight unmet safety needs. Experts call for tighter regulation and international governance to manage AI's transformative risks.

PLA-linked Chinese institutions are utilizing AI model distillation, a technique to train smaller AI systems using outputs from OpenAI and Anthropic models, to bolster their defense capabilities. This approach allows for advanced AI training on limited hardware, raising concerns about the strategic implications for U.S. tech companies and the potential deployment of these distilled models in military applications.

CRISPR technology has been used to create two hypoallergenic beagles, Bailey and Alfie, by suppressing the Can f 1 allergy protein. This breakthrough by Kindred Companion Sciences could transform pet ownership for allergy sufferers, though it raises ethical concerns and questions about adoption dynamics.

River Mobility, an Indian EV startup, has announced a $120 million Series C funding round led by Yamah Motor Co to scale its manufacturing and expand its product lineup. The company, competing with Ather and Bajaj, aims to boost production capacity and launch new models by 2028, driven by their current success with the Indie scooter.

Pokopia's 2.0 patch delivers substantial updates, featuring new gameplay mechanics, expanded customization, and significant bug fixes. Since its March release, the game has sold 4 million units, reinforcing Nintendo's lineup amid Switch 2's market challenges.

Representative Rashida Tlaib introduces a bill to block AI data centers on federal lands, citing pollution, noise, and potential 267% utility cost hikes. The initiative highlights environmental and community concerns amid defense plans for AI-enabled data centers, which may bypass regulations.

Waze's 2026 update for Apple CarPlay and Android Auto emphasizes personalization and safety with customizable voices and smarter route learning. Enhanced voice interaction and alerts aim to make driving safer and more intuitive.

The U.S. 9th Circuit Court of Appeals ruled against Amazon's claim that Perplexity’s AI agents violated federal computer-hacking laws. The decision, a landmark for AI legality, allows Perplexity's tools on Amazon, emphasizing user-driven access rather than independent AI actions.

Delhi High Court orders removal of infringing content targeting Physics Wallah founder Alakh Pandey to protect his personality rights. Platforms like Google and Telegram must share subscriber info to aid legal action.

MacPaw and Liquid AI have teamed up to develop an on-device AI stack for Macs, leveraging Apple Silicon's capabilities for fast, private AI operations. The partnership focuses on providing a comprehensive AI solution with both local and cloud processing options, prioritizing user privacy and persistent memory functionality.

Russia's largest-ever VPN crackdown began on August 4, 2026, targeting over 20 popular services, including Paper VPN and Amnezia VPN, with sweeping network restrictions. Some users report restored connectivity as providers adapt, while speculation arises about a potential state-run VPN solution.

A House Select Committee report reveals Chinese telecom giants are still deeply embedded in U.S. infrastructure, posing significant security threats. Despite losing FCC licenses, these carriers maintain a physical footprint, prompting urgent calls for policy changes, including enhanced oversight and potential removal of risky infrastructure.

Tencent debuts its Hy3 language model globally through platforms like WorkBuddy, offering free access until August 31, 2026, aiming to compete with Western models. Hy3 boasts advanced task execution and affordability, rapidly gaining global traction and reshaping the AI landscape with its open-source approach and extensive deployment capabilities.