Author: Art Ryan

AI had one of those weeks where the loudest story was not a single product launch. It was the pressure building underneath everything: open-weight models getting stronger, enterprises asking harder ROI questions, regulators circling real-world risks, and healthcare and education groups trying to turn AI into something useful outside the demo room. Technology & Innovation The biggest model story of the week came from China. Moonshot AI unveiled Kimi K3 on July 17, positioning it as a 2.8 trillion-parameter open-weight model and one of China’s strongest attempts yet to close the gap with leading US labs. The model drew attention…

China’s AI race just got louder. Moonshot AI, one of China’s most closely watched artificial intelligence startups, has released Kimi K3, a new open-weight model that it says can compete with some of the strongest systems from OpenAI and Anthropic. The claim is bold, of course. AI companies love bold claims. But this one is getting attention because Kimi K3 is not being positioned as a small experimental model or a cheap local alternative. It is being pitched as a serious frontier-level system from China. According to reports, Kimi K3 comes with 2.8 trillion parameters, making it one of the…

Roblox is making game creation feel a lot less like traditional development and a lot more like typing an idea into your phone. The company has introduced Build, a new AI-powered creation feature inside the Roblox mobile app. The basic idea is simple: users describe the kind of game they want, and Roblox’s AI tools generate a starting version that can be edited, tested, shared, and eventually published. That sounds small at first. It is not. For Roblox, this could turn millions of players into casual creators. For the wider gaming industry, it raises a bigger question that keeps coming…

Artificial intelligence may be heading further into the field. Literally. A new bipartisan proposal known as the FARM AI Act is drawing attention because it would bring more artificial intelligence support into U.S. agriculture, especially through research funding, farmer education, and workforce training. The idea is simple enough on the surface. Farms are getting more technical. Equipment is getting smarter. Data is becoming more important. But many producers still do not have the money, training, or local support needed to actually use AI tools in their daily operations. That is the gap the bill is trying to close. The FARM…

A strange new Claude model name briefly appeared in Cursor, sparking widespread speculation about a Claude Honeycomb leak, and the AI crowd did what the AI crowd always does. It took screenshots. Theories appeared almost immediately. Then the guessing started over what Anthropic might be preparing next. The model was reportedly listed as Claude Honeycomb EAP, and its short appearance has now triggered speculation that Anthropic may be testing something connected to its next major Claude release. Some users are calling it an early sign of Claude Opus 5. Others are being more careful, because right now, that part is…

Indonesia is trying to draw a line that many countries are still dancing around. Its draft copyright law says artificial intelligence can help create content, but it should not be treated as the author. That sounds simple. It is not. The question of who owns AI-assisted work has become one of the messiest legal fights in the creative economy. The proposal could make Indonesia one of the first countries in Southeast Asia to put AI directly into copyright law instead of leaving the issue to courts, policy notes, or vague platform rules. The main idea is clear enough: copyright belongs…

Google’s next major AI model is not arriving as quickly as expected. The company has reportedly delayed the launch of Gemini 3.5 Pro by several months after internal testing showed that the model’s coding abilities were not where Google wanted them to be. That is a serious problem now. Coding is no longer a side feature for frontier AI models. It has become one of the main battlegrounds. According to Times of AI, the delay follows reports that Gemini 3.5 Pro fell short of internal standards, especially in software development tasks. The model had been expected after its introduction at…

Frontline workers make up most of the global workforce, yet many of them still sit outside the digital systems companies use every day. As businesses consider the impact of an AI frontline workforce, the digital divide becomes even more apparent. Many workers have no corporate email. HR portals are often difficult to access. A desktop computer is not waiting at the start of every shift. Instead, communication may depend on a phone, a manager, a paper form, or a WhatsApp message that may or may not reach the right person. Humand is trying to close that gap with an AI-powered…

The United States has made a major move in its technology relationship with the United Arab Emirates, easing export controls and giving the Gulf country wider access to advanced AI chips and other strategic technologies. This is not a small paperwork adjustment. The US Commerce Department has moved the UAE into Country Group A:5 under the Export Administration Regulations, a trusted export category that gives approved partners easier access to sensitive American technologies. At the same time, the UAE has been removed from more restrictive categories, including D:3 and D:4. That matters because those classifications affect how difficult it is…

Saudi Arabia is bringing artificial intelligence deeper into its cultural economy. The Kingdom’s Ministry of Culture and the Cultural Development Fund have signed a trilateral memorandum of understanding with the Saudi Data and Artificial Intelligence Authority, better known as SDAIA. The agreement is designed to explore AI programmes that can support creators, cultural enterprises, and institutions across Saudi Arabia’s officially recognised cultural sectors. It is not the usual AI story about banking, healthcare, cloud infrastructure, or data centers. This one moves into museums, heritage, fashion, film, music, theatre, literature, culinary arts, and other creative fields that are now being treated…

OpenAI wants AI rules in the United States to stop looking like a messy state-by-state experiment. The company is pushing for a single federal framework for frontier AI, one that could override some state laws already moving through places like California, New York, Illinois, and Massachusetts. That sounds like normal policy language. It is not. This is one of the biggest fights in AI regulation right now: who gets to write the rules, Washington or the states? OpenAI is calling its idea “reverse federalism.” Basically, states can test early laws, Congress can take what works, and then the federal government…

Thinking Machines has finally shown what it has been building. The AI startup founded by former OpenAI CTO Mira Murati has launched Inkling, its first major AI model, and the message is pretty clear. This is not another model release trying to win the internet for a day with leaderboard screenshots. Inkling is being positioned around something less flashy but probably more important for businesses: control. For enterprises, that word matters. A lot. Most companies using AI today are still renting intelligence from someone else. They plug into a closed API, send prompts, wait for responses, and hope the model…

AI is no longer sitting politely inside office software. Applied Computing oil plant AI is now transforming industries beyond the office, including complex environments like oil plants. It is moving into refineries, petrochemical sites, and oil and gas plants where pressure, heat, equipment failure, and production delays are not small problems. That is the space Applied Computing wants to enter more aggressively after raising $20 million in Series A funding for its industrial AI platform. The London-based startup is building Orbital, a foundation AI model designed for energy and industrial plants. Not a general chatbot. Not a dashboard with a…

The country is moving deeper into sovereign AI, with Japanese universities, startups, telecom players, manufacturers, and major enterprises building localized AI systems on top of NVIDIA Nemotron open models. The goal is not just to use smarter chatbots. Japan wants AI that understands Japanese language, Japanese industries, Japanese regulations, and Japan’s very real workforce problem. That last part matters. Japan’s aging population and shrinking labor force are not abstract policy issues. They are already reshaping factories, hospitals, telecom networks, public services, customer support, and enterprise operations. So the country’s AI strategy has a practical edge: build systems that can support…

Dubai based AI & communications strategist Shadi Dawi has earned his Doctorate of Business Administration with Premium Distinction from ESC Clermont School of Business in France, concluding two years of intensive research and preparation that ended in what jury members called one of the strongest defences the programme has witnessed. His dissertation, titled “AI Truth”: A Framework for Governing AI Generated Content in UAE Corporate Communications, confronts a problem most organizations now live with but few have solved. Generative AI drafts press releases, executive statements, and crisis communications at a pace no editorial process was designed to match. Production has…

AJMS Group and Marmin AI launch the CXO Boardroom Series in Dubai, a curated executive forum on AI, digital finance, and enterprise transformation, managed by Trescon and opening 22 July 2026 AJMS Group, one of the region’s most established business advisory groups, and Marmin AI, its AI-powered compliance and finance technology entity and the first fully approved e-invoicing Accredited Service Provider (ASP) in the UAE, operating with 100% local data residency, have announced the launch of the CXO Boardroom Series, a curated invitation-only programme for senior executives across the UAE and GCC. Managed end-to-end by Trescon, a leading global business…

Perplexity has released a new benchmark called WANDR, and it quietly says something many people using AI research tools already know. These agents can be impressive. They can move fast. They can find sources, summarize pages, compare companies, build lists, and produce something that looks useful. Then the hard part arrives. Can they do it again? And again? Across dozens of companies, people, filings, dates, sources, and claims, without dropping pieces along the way? That is where Perplexity’s WANDR benchmark starts asking uncomfortable questions. What Is Perplexity WANDR? WANDR stands for Wide ANd Deep Research. It is an open benchmark…

IntelAgree is pushing its contract intelligence platform further into agentic AI with the launch of Saige Assist: Agent, a new general-purpose AI agent built for contract portfolios. This is not being positioned as a small chatbot feature attached to a contract platform. IntelAgree says the agent can work across a customer’s clause library, playbooks, negotiation history, and contract data to answer questions, compare versions, redline agreements, build dashboards, run analysis, and even make approval-gated updates inside the CLM platform. That last part matters. A lot of AI tools in legal tech still stop at “here is an answer.” Useful, sometimes.…

OpenAI may be getting ready to take ChatGPT off the screen and place it somewhere much more personal: inside the home. The company is reportedly working on an OpenAI smart speaker powered by ChatGPT, a device that could put OpenAI directly against Amazon Alexa, Google Assistant, and Apple’s HomePod. This is not just another rumor about AI hardware. It feels more like OpenAI testing whether people actually want an AI assistant that lives with them, listens better, understands context, and does more than answer basic commands. For years, smart speakers have mostly been useful in a limited way. Play music.…

Moonshot AI has not officially launched Kimi K3 yet, but the internet is already acting like it has. A leaked page and a wave of community reports have pushed Kimi K3 into the center of AI chatter, with claims that the next-generation model could bring a massive jump in scale, context length, coding ability, and agent-style workflows. The careful part first: most of this is still unconfirmed. Moonshot AI has not released a full model card, technical report, pricing page, benchmark sheet, or official launch statement for Kimi K3. So the details floating around right now should be treated as…