Anthropic nears $20B funding at $350B valuation, doubling target due to strong investor demand. Key backers include Nvidia and Microsoft, with recent product successes boosting its appeal. Rival OpenAI also raising funds, with both eyeing summer IPOs.

Workday CEO Carl Eschenbach steps down immediately, succeeded by co-founder Aneel Bhusri, who returns as permanent CEO to lead the company's AI-focused next phase.

OpenAI launches ChatGPT ads for US users to offset infrastructure costs. Ads target adult free-tier users with privacy safeguards and opt-out options.

Claude Opus 4.6 leads the AI benchmark, excelling in agent workflows and coding, though testing costs were high due to its token pricing structure.

Anthropic's new "Fast Mode" for Claude Opus 4.6 costs 6x more but is 2.5x faster, ideal for real-time debugging. Available via Claude Code with a 50% discount until Feb 16, it's not on major cloud platforms.

Salesforce shifts focus from Heroku PaaS, halting new enterprise sales and scaling back feature development while maintaining core services and security updates.

Apple is integrating third-party chatbots like ChatGPT, Claude, and Gemini into CarPlay, allowing users to manually launch them for voice control, with a potential rollout within months.

Crypto.com founder buys AI.com for $70M in crypto, setting a new record for the most expensive domain purchase. The site will launch with personal AI agents for messaging, apps, and trading.

Mistral AI launches Voxtral Transcribe 2, a competitively priced speech recognition model family starting at $0.003/min, offering real-time and batch processing variants with 13-language support.

Nvidia cancels RTX 50 Super launch to prioritize AI chip production amid memory shortages, delaying next-gen gaming GPUs and reducing RTX 40 series output.

Higgsfield launches Vibe Motion, a no-code AI tool for creating dynamic graphics and animations in minutes, targeting marketing and presentations with plans from $17.40/month.

Meta is testing a standalone app, Vibes, for creating and sharing AI-generated short videos. It functions like an AI-only TikTok, separate from Meta AI, to better compete with OpenAI's Sora. A freemium subscription model for advanced features is planned.

OpenAI launches Frontier platform for AI agent creation and management, alongside GPT-5.3-Codex model for enhanced coding and productivity.

Anthropic launches Claude Opus 4.6 with a 1M token context window, claiming mitigated performance decay. It outperforms GPT-5.2 on key benchmarks, introduces new API features like adaptive reasoning and compression, and enhances office integrations. However, it shows increased vulnerability to indirect prompt injections.

Amazon's MGM Studios AI tools for film and TV production are moving to closed testing in March, with initial results expected in May. The initiative, involving industry partners, aims to boost efficiency and cut costs while protecting IP, though AI adoption in Hollywood sparks job and creativity debates.

Google's parent Alphabet exceeds $400 billion in annual revenue for the first time, driven by cloud and YouTube growth. YouTube revenue surpasses $60 billion, while cloud hits $70 billion run rate. Gemini AI users top 750 million, and AI search queries double.

GitHub integrates Anthropic's Claude and OpenAI's Codex AI agents into its platform, available in public preview for Copilot Pro Plus/Enterprise users. Part of "Agent HQ," developers can now choose between Copilot, Claude, Codex, or custom agents for tasks, using premium requests per agent. They can also evaluate each agent's performance. GitHub is rapidly adopting competing AI models to enhance Copilot, with plans to expand access to more subscription tiers and collaborate with Google, Cognitio

Anthropic pledges to keep its AI chatbot Claude ad-free, contrasting OpenAI's plans for ChatGPT ads.

OpenAI hires former Anthropic AI safety expert Dylan Scandinaro as Head of Emergency Preparedness to enhance safety for powerful AI models amid high-risk evaluations.

Alibaba launches Qwen3-Coder-Next, an 80B-parameter open-source AI model for coding, claiming superior performance on benchmarks with a 256K token context.