In a day filled with significant developments in the AI landscape, OpenAI’s strategic moves and Google’s Gemini app’s rapid growth stand out.

Xiaomi’s MiLM Plus Releases PROVE Metrics for Object Removal

Xiaomi has introduced two new metrics, RC-S and RC-T, for evaluating object removal models, addressing the growing gap between model performance and assessment tools. These metrics aim to provide a more accurate evaluation of how well these models reconstruct complex visual elements like shadows and reflections, which traditional metrics often fail to assess correctly.

OpenAI Lets Employees Cash Out $7 Billion in Stock

OpenAI has completed a substantial $7 billion stock buyback, allowing both current and former employees to sell shares at a valuation of $852 billion. This initiative is designed to alleviate liquidity pressures on employees as the company prepares for a potential IPO, following a similar buyback of $6.6 billion in late 2025.

Brad Lightcap, OpenAI’s COO, Announces Departure

In a significant leadership change, Brad Lightcap, OpenAI’s long-serving COO, is leaving the company to pursue new ventures. His departure marks a notable shift in the executive team, although he expressed enthusiasm about continuing to support the company’s mission from a different perspective.

Gemini Becomes Google’s Fastest-Growing Product with 1 Billion Users

Google’s Gemini app has achieved a remarkable milestone, surpassing 1 billion users, making it the fastest-growing product in the company’s history. The app’s success is underscored by its user engagement, with a significant portion utilizing the voice feature to interact with the AI assistant and generating over 150 million images daily.

General Catalyst Leads $1.1 Billion Investment in River AI

River AI, a new startup co-founded by Igor Babuschkin of xAI, has secured an impressive $1.1 billion in funding just two months after its inception. This substantial investment highlights the growing interest in personal AI agents and the potential impact of River AI’s innovative vision in the field.

Security Researchers Uncover Vulnerabilities in AI APIs

A recent investigation has revealed vulnerabilities in the APIs of major AI companies, including OpenAI and Google. Researchers discovered that these weaknesses could allow unauthorized access to encrypted reasoning traces, exposing sensitive information such as passwords and API keys, raising significant security concerns for users.

Unreleased Anthropic Model Advances Riemann Hypothesis Research

In a notable development, an unreleased model from Anthropic has made unexpected progress on the Riemann hypothesis, one of mathematics’ most significant unsolved problems. While the model has not provided a solution, its advancements demonstrate the potential of AI in tackling complex mathematical challenges.

Nvidia’s Nemotron 3.5 Lightning Prioritizes Speed

Nvidia has launched the Nemotron 3.5 Lightning, an open-weight AI model that emphasizes speed and efficiency over sheer intelligence. Despite having fewer parameters compared to larger models, it boasts impressive processing capabilities, showcasing Nvidia’s strategic focus on creating faster, more efficient AI solutions.

Spotify Introduces AI Persona Labels for Artist Profiles

Spotify is rolling out new labels for AI-generated artist profiles, termed “AI Persona,” which will exclude their music from personalized recommendations by default. This move aims to clarify the distinction between human and AI-created content within the platform, reflecting the increasing prevalence of AI in the music industry.

Compiled automatically by the Tech AI Newsdesk from public AI-news sources and summarised in our own words.