Meta knowingly took in billions from scam ads on Facebook and Instagram, says Reuters

Meta is developing a new AI model focused on images and videos, codenamed Mango, which is expected to be released in the first half of 2026 alongside a new language model named Avocado. This initiative is part of Meta's ongoing efforts to enhance its artificial intelligence capabilities, as stated by Alexandr Wang, the company's AI chief.

James Proud, founder and CEO of Substrate, emphasized the need for the U.S. to adopt a 'leap-ahead' strategy in the AI race against China during an interview on Bloomberg Tech. He highlighted the increasing competition in artificial intelligence between China and the West, urging for proactive measures to maintain technological leadership.

Palona AI, a startup founded by former Google and Meta engineers, has launched Palona Vision and Palona Workflow, marking a significant shift towards the restaurant and hospitality sectors. These new features aim to create a real-time operating system for restaurant operations, integrating various modalities such as cameras and task execution. This launch represents a strategic pivot from the company's initial focus when it debuted in early 2025 with $10 million in seed funding.

China is advancing its efforts toward chip self-reliance by testing a secret extreme ultraviolet (EUV) lithography prototype, a critical technology for semiconductor manufacturing, amid increasing export controls and rising demand for artificial intelligence (AI) technologies.

China has reportedly reverse-engineered an extreme ultraviolet (EUV) lithography machine, a significant advancement in its semiconductor manufacturing capabilities. The project, completed in early 2025, involved a team of former ASML engineers and marks a pivotal step toward achieving chip self-sufficiency in the country.

Informal classes conducted via WhatsApp and Facebook groups are empowering freelancers and business owners in India to acquire essential AI skills, enhancing their capabilities in a rapidly evolving digital landscape.

The study on sentiment analysis during Bangladesh's 2024 mass uprising highlights the emotional dynamics of civil unrest, focusing on public sentiments categorized as Outrage, Hope, and Despair. Utilizing a dataset of 2,028 annotated news headlines from Facebook, the research employed Latent Dirichlet Allocation (LDA) to identify themes such as political corruption and the impact of internet blackouts on sentiment patterns.

Recent research has leveraged Large Language Models (LLMs) to effectively detect and respond to online harassment on Instagram, focusing on private messaging rather than public social media platforms. Human labelers were recruited to identify harassment in a dataset of Instagram messages, and the LLM pipeline demonstrated its capability to label these messages accurately while generating superior simulated responses to harassment.