Tech Research Online

In Focus Kim K3 is the largest AI model in China with 2.8 trillion parameters The model performed better than GPT-5.5 and Opus 4.8 in general agents and coding Kimi K3 will cost more than comparable Chinese open-weight models Moonshot AI has launched a new model which it claims performs better than OpenAI and Anthropic’s most advanced models on some benchmarks. With 2.8 trillion parameters, Moonshot’s Kimi K3 is the largest AI model in China.

In Focus The order required Google to give rival AI assistants access to 11 Android features The order falls under EU’s Digital Markets Act The tech giant must also share anonymized search data with AI chatbot providers The EU has ordered Google to share data and open up key features of its Android operating system to OpenAI and rival AI services.

In Focus The AI Mode update allows users to connect to preferred apps directly in Search Users can complete tasks across the apps they regularly use The AI Mode third-party app update is rolling out to users across the U.S. Google has updated AI Mode to enable it to support third-party apps. The Google AI Mode connected apps update allows users to link to their preferred apps directly in Search.

In Focus Meta’s safety system flags conversations with self-harm statements Human reviewers check conversations before alerting parents Meta AI parental alerts have rolled out in the U.K., the U.S., and Canada Meta AI will now alert parents when their teens discuss self-harm or suicide with the chatbot. Meta also plans developing emergency response capabilities that would allow Meta AI to connect users with relevant services where conversations suggest an immediate safety risk.

In Focus GPT-Red completes red-teaming tasks automatically OpenAI has integrated GPT-Red directly into GPT-5.6’s training process GPT-Red delivers 6x less direct prompt injection failures than GPT-5.5 OpenAI has unveiled a cybersecurity-focused AI model called GPT-Red. As AI systems become major targets of cyber attacks, OpenAI is using GPT-Red to protect GPT-5.6 against prompt-injection attacks.

In Focus Microsoft wants its sales teams to be more competitive with OpenAI and Anthropic The company asked salespeople to position its AI offering as an end-to-end solution CEO Satya Nadella emphasized the need to help enterprises to optimize AI-related costs Microsoft is reportedly training its salespeople to get more competitive with OpenAI and Anthropic AI. The software giant is asking employees to position its AI offering as an end-to-end solution.

Introduction The global subscription economy is projected to reach $859 billion by the end of 2026. Nearly every digital interaction from watching television and managing passwords to tracking workouts and generating code, is gated behind a recurring monthly fee. For years, digital platforms relied on a simple premise: if the friction to sign up is low, consumers will set up auto-pay and forget about the charge. That assumption is no longer safe.

In Focus Anthropic, Blackstone and Hellman & Friedman announced the formation of Ode this year The company designs custom systems for individual organizations Ode has a workforce of 100 engineers Anthropic, Blackstone, Hellman & Friedman, Goldman Sachs and other partners have launched Ode with Anthropic to fast-track enterprise AI adoption.

In Focus Anthropic featured facial surveillance and graveyards in its latest ad OpenAI CEO Sam Altman said he thought the ad was satire The cemetery imagery in the ad attracted significant criticism Anthropic has faced backlash over a new ad titled “There’s Hope in Hard Questions”. The ad opens with scenes of burning houses, homeless individuals, vast graveyards and facial surveillance. A voiceover questions whether AI can be trusted and who intervenes when the system goes off the rails.

In Focus OpenAI’s hardware device will reportedly be a mobile, screenless smart speaker The smart speaker will rely on ChatGPT Voice Mode for communication OpenAI envisions a device that can anticipate user needs and retrieve information proactively OpenAI’s first hardware device will reportedly be a mobile, screen-free smart speaker. The new device is designed to be a new type of home computer and is expected to serve as a companion device in homes.