OpenAI releases GPT-5.2 Codex model via API, excelling in complex coding tasks and cybersecurity. Priced at $1.75/M input tokens, it supports text/image inputs with four reasoning levels. Integrated by Cursor and Windsurf (limited-time discount).

Phenom People acquires AI analytics platform Included to enhance workforce intelligence and decision-making.

Tesla shifts FSD to subscription-only model, eliminating one-time purchase option. Move aims to lower access barrier amid intensifying regulatory scrutiny and BYD surpassing Tesla in 2025 EV sales.

Google's Gemini app introduces "Personal Intelligence," an experimental feature that personalizes AI responses by connecting with your Google data (Gmail, Photos, Search, YouTube). It reasons across this data for proactive, context-aware answers. The feature is opt-in, with safeguards for sensitive topics, and is initially available to US subscribers before a wider rollout.

OpenAI partners with Cerebras in a multi-year deal for 750MW of AI computing power through 2028, valued over $10 billion, to accelerate customer AI inference and responses.

Apple launches Creator Studio subscription service offering Final Cut Pro, Logic Pro, Pixelmator Pro, and premium iWork content for $12.99/month.

Instagram co-founder Mike Krieger is shifting to co-lead Anthropic's internal "Labs" incubator, focusing on experimental AI products, as the company expands the team. This executive move comes amid intense competition in the AI industry, with giants and startups vying for dominance and raising massive funding.

Microsoft unveils a five-point plan to address community concerns over data centers, focusing on preventing electricity rate hikes, reducing water use, and increasing transparency amid growing local pushback.

Meta's Reality Labs to lay off 10% of its workforce as it shifts investment focus from the metaverse to wearables, aiming to boost smart glasses production by late 2026. The metaverse strategy will pivot toward mobile accessibility.

Google's Veo 3.1 update enhances video generation with improved "Video Elements" based on reference images, adds native vertical video support for platforms like TikTok, integrates features into Gemini and YouTube apps, and introduces 4K upscaling from 1080p.

Salesforce launches enhanced Slackbot AI, now powered by Claude LLMs. It automates info retrieval from docs and apps, generates content, and will integrate with AI agents. Available in Business+ and Enterprise+ plans.

UK law criminalizes non-consensual deepfake creation, effective immediately. Ofcom investigates X's Grok AI for potential Online Safety Act violations, with fines up to £18 million or 10% of global revenue. X restricts Grok's image generation to paid users, but free workarounds persist.

Anthropic launches Cowork, a low-barrier Claude Code alternative. It lets users manage files via chat, targeting non-coders. Currently in preview for Max subscribers.

Meta launches Meta Compute to boost AI infrastructure, planning to scale energy capacity to hundreds of gigawatts. Key executives will lead technical, strategic, and government collaboration efforts.

Anthropic launched Claude for Healthcare, a specialized AI tool for medical data integration and administrative automation, while cautioning users to consult professionals for medical advice.

Amazon introduces Bee AI wearable at CES, a pin/bracelet device for recording conversations and acting as a personal AI companion. It complements Alexa, with future integration planned to enhance user experience across home and outside environments.

OpenAI acquires health tech startup Torch Health to enhance AI-driven medical data integration and accelerate ChatGPT Health development.

Apple partners with Google, integrating multiple Gemini AI models into Siri for enhanced personalization and multi-step task handling, while maintaining its existing OpenAI collaboration.

DeepSeek researchers introduced multi-stream constrained HyperConnections (mHC) to stabilize training in large language models while enhancing performance through optimized connectivity.

Lightricks open-sources LTX-2, a 19B-parameter AI model for synchronized audiovisual generation, offering superior speed, multimodal coherence, and 20-second high-fidelity outputs; supports 4K, 50fps, real-time lip-sync, and runs on consumer GPUs.