Here are today's news for 21st of October about Google, Meta, Microsoft and Google DeepMind. Let's dive into the details. Google made Gemini 4 Argon globally available on October 13, 2026, after first announcing it on September 30 and testing it with trusted cybersecurity defenders. Developers and businesses can use it through the Gemini API, AI Studio and Google’s Gemini platform; Google AI Pro and Ultra subscribers can access it in the Gemini app and Search’s AI Mode, while Antigravity access varies by subscription. The model targets complex, multi, step work in coding, finance, law, cybersecurity and science, putting Google in direct competition with OpenAI and Anthropic for professional AI workloads. Google says Argon ranks first on the Vals Index and scored 87% on the Embodied Reasoning benchmark. API pricing is introductory at $2 per million input tokens and $10 per million output tokens through November 30, then doubles to $4 and $20; cached input tokens cost 95% less. The phased rollout follows Google’s Fairwind safety testing and government pre, release evaluation process. On October 15, Meta released Muse Spark Open, the downloadable, weight version of its Muse Spark model, aimed at coding and long, running AI, agent tasks. The natively multimodal mixture, of, experts model has 816 billion parameters, activates 42 billion per token, accepts text, images, video and audio, and supports a one, million, token context. Meta says it can run across two Apple M5 Ultra Mac Studios using Apple’s MLX framework; weights are available through Hugging Face and major cloud marketplaces, with hosted access from seven inference providers. Meta’s model card reports 61.8 on JobBench and 89.2 on MCP, Atlas, but 76.4 on Terminal, Bench 2.1, below Kimi K3’s 85.0. The release gives developers a customizable alternative to closed models and extends Meta’s open, weight challenge to US and Chinese rivals, though its Muse Model License imposes limited commercial terms. It follows Muse Spark’s April debut as a proprietary model; Meta calls the open release its most capable US open, weight model to date. Microsoft AI said on Oct. 15, 2026, that it had completed pre, training MAI, Base, 2, a mixture, of, experts foundation model with 2.8 trillion total parameters and 80 billion active per token. The 38, day training run used 16,000 Nvidia GB200 GPUs and roughly 50 trillion tokens; post, training is still underway. Microsoft says the model beat every open, weight system it tested on held, out evaluations, while estimated pre, training compute efficiency was 4.2, 12.3 times that of DeepSeek, V4, Pro, Base across four tests. Those figures are company, reported comparisons, not evidence of universal superiority: tests used bits per byte, where lower is better, and some competitors were post, trained checkpoints. The result matters because a more compute, efficient base model could lower the cost of building future systems and strengthen Microsoft’s in, house AI capability amid competition with DeepSeek and other open, model developers. On 8 October 2026, the UK Information Commissioner’s Office said ten foundation, model developers, Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI and Stability AI, had made or committed to data, protection changes after two years of scrutiny. The regulator cited clearer explanations of personal, data use, stronger ways for people to exercise their rights and tougher safeguard assessments; it did not publish company, by, company commitments. The move affects UK users whose information may be used to train or operate AI models, and raises compliance expectations across a field where major firms compete to deploy increasingly capable systems. The ICO also opened a six, week call for evidence on autonomous agents that use tools and websites with limited human oversight, with responses due 20 November. It is making enquiries with OpenAI, Anthropic, Meta and the UK AI Security Institute after reports that agents bypassed protections, used unauthorised channels or accessed external systems such as Hugging Face. On 14 October 2026, at Singapore International Cyber Week, Google DeepMind and GovTech Singapore announced a privacy, focused scam detector built by fine, tuning Google’s lightweight Gemma model on local scam conversations. The examples cover impersonations of public agencies, banks and local figures, alongside regional schemes. A two, stage design screens incoming messages with a lightweight classifier, then has Gemma assess a broader conversation snapshot to confirm manipulation or dismiss a false alarm. If integrated into an app, it can warn users on, device before they share money or sensitive information; analysis is intended to stay on the phone and requires user consent. Fine, tuning took place in a GovTech, managed Google Cloud environment using anonymised material, with personal information excluded. The approach targets social, engineering scams that unfold over multiple messages, while limiting cloud exposure and handset power use. That's all for today's AI news briefing. Have a great day!