Generative AI

A Singaporean man was left traumatized after discovering AI-generated images that falsely depicted him in a family with a former schoolmate. The incident highlights the emotional harm of non-consensual synthetic media.

Google has launched Ask Ad Manager, a Gemini-powered AI assistant that lets advertisers and publishers ask questions to troubleshoot campaigns and get personalized guidance directly inside Google Ad Manager.

President Trump declared Anthropic no longer a national security threat, signaling an end to the 10-day export control freeze. The reversal follows a competitor's tip and a collapsed Pentagon contract over surveillance and weapons use.

Apple announced Core AI, a framework that runs generative AI models up to 70 billion parameters entirely on-device. It succeeds Core ML and powers Apple Intelligence with a unified API across Apple silicon.

Zhipu AI's market value crossed HK$1 trillion as shares jumped 42% on Monday. The surge followed the GLM-5.2 model release, with founder Tang Jie challenging Elon Musk's timeline for Chinese AI dominance.

Some Asia hedge funds delivered returns exceeding 100% in early 2026, fueled by AI hardware and large language model bets. The rally pushed Japan, South Korea, and Taiwan indexes to record highs despite geopolitical tensions.

VeloAlpha, a Beijing-based start-up, launched FusionAlpha in April to tackle fusion energy's simulation bottleneck. The AI-driven tool promises to break the 'impossible triangle' of speed, accuracy, and physical fidelity, potentially accelerating commercial fusion development.

Snap is turning its internal AI video generation team into a standalone company named Dotmo, focusing on interactive gaming. The move aims to manage high development costs while retaining equity upside.

A brutal assault in Hachioji reveals how Japanese teenagers are using AI to plan violent acts, sparking a national reckoning over digital dependency and eroded autonomy. Experts warn this crisis could be a global harbinger.

Norway will ban generative AI in elementary classrooms and restrict it in higher education starting August 2026. Prime Minister Støre says the move protects foundational learning and cognitive development.

A U.S. directive forced Anthropic to block foreign access to its most advanced AI models, revealing a dangerous vacuum in international oversight. The incident has intensified demands for transparent, binding global rules on frontier AI deployment.

Norway will ban generative AI for elementary school pupils starting in August 2026, while allowing supervised use for older students. The policy reverses decades of digital classroom adoption to protect foundational learning.

Researchers found that the prompt "restore the attached photo" caused ChatGPT to generate graphic, sexual, and violent images without any image attached. OpenAI has implemented new safeguards, but the incident highlights ongoing AI safety challenges.

Google’s Gemini Live voice assistant can now recall past conversations and access connected apps for personalized responses. The update brings voice interactions closer to the text-based Gemini experience, though some app integrations remain unavailable.

Adobe expanded Firefly with AI tools for brand kits, storyboards, and video production, plus a preview of a unified creative studio now in private beta. The creative AI agent also launched across Premiere Pro, Photoshop, and more.

Chinese AI firms are rapidly deploying trillion-parameter models, driven by lower costs and domestic chip adaptation. The move intensifies the US-China tech rivalry amid fresh export curbs.

Anthropic's most advanced AI, Mythos Preview, remains accessible to a chosen few despite a U.S. government ban. The selective access raises transparency and security questions.

At Reliance's 49th AGM, Mukesh Ambani faces investors with a Jio IPO potentially worth $4 billion, a new Meta AI data center in Jamnagar, and green energy updates. The meeting could reshape India's AI and energy landscape.

ChatGPT's health intelligence upgrade delivers precise, citation-backed answers on medical decisions. The model now explains why an MRI before a sciatica steroid injection is crucial for safety and efficacy.

Dataland opens as the first AI arts museum, using a foundational model and Google Cloud to create a responsive, multi-sensory rainforest exhibit. The project pushes boundaries but raises questions about transparency and artistic integrity.

Google has rolled out Gemini-powered email summaries to free Gmail users globally, but the feature can be disabled—at the cost of losing other AI tools. Here's what you need to know about privacy and control.

Google's June Pixel Drop introduces Gemini Omni 2 for AI video creation, Screen Reactions for creators, and floating app Bubbles for multitasking. The update also expands Voice Translate, safety alerts, and Edit with Ask Photos to more devices and regions.

LinkedIn and Adobe have launched free, role-based AI training for marketers on LinkedIn Learning, available in 47 languages. The program responds to a 113% increase in AI-related marketing job postings and offers certificates to boost professional profiles.

AWS Marketplace introduced an AI tool that automatically creates product listings from partners’ existing digital assets like websites and PDFs. The assistant validates content, optimizes for search, and gives a quality score to boost buyer discoverability.

Google's June Pixel Drop debuts Screen reactions for easy reaction videos, plus Gemini upgrades and Android 17. The rollout starts today, targeting creators and power users.

AWS invented P-EAGLE to eliminate sequential drafting in speculative decoding, achieving up to 1.69x throughput gains. The method is now natively available in Amazon SageMaker JumpStart for one-click deployment.

Carry on Jatta 4 has digitally brought back late actor Jaswinder Bhalla using AI, with his son performing on set and Gippy Grewal dubbing. The family-approved move raises new ethical questions about AI in filmmaking.

AWS details a generative AI pipeline that combines Bedrock Data Automation, Step Functions, and agentic coordination to extract insights from complex documents. The architecture reduced real estate report screening from hours to minutes and scaled to 50,000 concurrent PDFs.