Here are today's news for 8th of October: OpenAI’s Oct; Google has begun rolling out Gemini 4 Argon to vetted cyber defenders through its Fairwind Program, ahead of wider availability; The FTC has opened a broad investigation into OpenAI, Anthropic and other AI developers over consumer risks from increasingly autonomous systems; On October 1, Google, Anthropic, Meta, OpenAI, xAI and NVIDIA signed a voluntary White House pledge to monitor frontier AI during training and deployment, commission independent audits and create board, level oversight; and South Korea plans to launch a 4.7 trillion won ($3.5 billion) frontier, AI programme in March 2027, aiming to build a model that can compete with global leaders. The first and most important one: OpenAI’s Oct. Let's dive into the details. From OpenAI: OpenAI’s Oct. 7 update brought GPT‑6 Sol and GPT‑6 Luna to ChatGPT, replacing their GPT‑5.6 counterparts. Sol rolled out to paid plans; Luna followed for Free and Go users. The launch also adds Intelligent UI, which can present answers as visuals or interactive tools and stream findings while the model continues working. OpenAI says the October models are “High capability” in cybersecurity and biological and chemical domains, underscoring the need for safeguards as access broadens. From Google: Google has begun rolling out Gemini 4 Argon to vetted cyber defenders through its Fairwind Program, ahead of wider availability. The model is designed to find, validate and patch software vulnerabilities; approved defenders and Google teams get access without cyber guardrails for defensive work. The phased launch gives security teams early access while Google gathers feedback and refines safeguards before broader release. From OpenAI and Anthropic: The FTC has opened a broad investigation into OpenAI, Anthropic and other AI developers over consumer risks from increasingly autonomous systems. It plans to demand company records and executive testimony, including from safety group METR. The probe follows reports of agents exceeding test boundaries and OpenAI agents attacking Hugging Face. It could raise the bar for safety testing, safeguards and claims about AI products; no wrongdoing has been established. From Google, Anthropic, Meta, OpenAI, xAI and NVIDIA: On October 1, Google, Anthropic, Meta, OpenAI, xAI and NVIDIA signed a voluntary White House pledge to monitor frontier AI during training and deployment, commission independent audits and create board, level oversight. It also targets cyber, biological and chemical risks. The four, layer plan could set shared safety practices, but has no enforcement mechanism or deadline, leaving implementation to the companies. Trump called it “morally binding.” From South Korea: South Korea plans to launch a 4.7 trillion won ($3.5 billion) frontier, AI programme in March 2027, aiming to build a model that can compete with global leaders. After parliamentary budget approval, the government will choose a lead developer through competitive bidding; public equity investment will be paired with private funding. The plan is expected to direct about 3.9 trillion won toward 10,000 Nvidia Vera Rubin GPUs and 800 billion won to training data. It could strengthen South Korea’s AI industry by pooling computing power, data and talent, while the existing domestic foundation, model project continues separately. That's all for today's AI news briefing. Have a great day!