AI Weekly Update - January 2026

OpenAI launches ChatGPT Go with higher limits and ad integration.

OpenAI has launched its ChatGPT Go globally, offering features such as GPT-5.2 Instant and increased memory limits. This new offering includes plans for ad integration, marking a notable shift towards monetizing free and lower-cost tiers of the ChatGPT service. This could lead to expanded user access while addressing ongoing high operational costs. The model is aimed at enhancing affordability and accessibility for users globally.

Anthropic expands Claude capabilities with new Cowork feature.

Anthropic has officially launched its Claude Cowork feature, which is now available for all users holding a $20 Pro subscription. This feature is designed to assist with collaborative tasks and streamline workflows. The expansion highlights the company's focus on enhancing user productivity, positioning Claude as a stronger competitor in the AI assistance space against other established players.

OpenAI collaborates with Cerebras to enhance AI compute resources.

OpenAI has partnered with Cerebras to add 750 MW of ultra low-latency AI compute to its platform, which is expected to significantly bolster processing capabilities. This partnership underscores the ongoing competition in the AI landscape, particularly in areas requiring substantial compute power for advanced AI tasks. Enhanced compute may offer OpenAI advantages in deploying more complex models at lower latencies.

New tariffs imposed on Nvidia's H200 AI chips headed to China.

Amidst rising geopolitical tensions, the US has imposed a 25% tariff on Nvidia's H200 AI chips destined for China. This move is seen as part of broader efforts to control technology transfer to China, reflecting national security concerns. The tariffs may impact Nvidia's operations and pricing strategies in international markets as the company navigates ongoing supply chain challenges.

Bandcamp bans AI-generated music, emphasizing human-created content.

In a significant policy shift, Bandcamp has become the first major platform to ban AI-generated music, citing a need to ensure authenticity and confidence in music production. This decision may inspire other platforms to reconsider their policies on AI content, reflecting a growing trend to prioritize human creativity in the music industry.

Internal reinforcement learning enhances LLM reasoning capabilities.

Researchers at Google have introduced an innovative method called internal reinforcement learning that aims to improve the reasoning abilities of large language models (LLMs). This technique shifts away from standard next-token prediction, guiding models to develop structured solutions for complex tasks. The strategy potentially reduces hallucinations in LLM outputs, which is pivotal as more advanced AI applications depend on reliable reasoning.

Google's Gemini model can now scan user photos and emails to improve responses.

Google has introduced a feature in its Gemini AI model allowing it to scan user emails and photos to provide more personalized responses. Initially available to paid users, this functionality marks a significant enhancement in AI interaction, promoting a more integrated user experience. By leveraging personal data responsibly, Gemini aims to deliver targeted and contextually relevant outputs, raising questions about privacy and data management practices.

OpenAI

Anthropic

Nvidia