Friday, 2 October 2026
This dated roundup collects the most interesting AI and technology developments found for Friday, 2 October 2026.
Policy & Ethics
AI Safety by Self-Policing Is Failing the Regulation Test
AI safety is being left to self-regulation, but critics say the U.S. needs enforceable rules, not voluntary pledges. The post AI Safety by Self-Policing Is Failing the Regulation Test appeared first on Superintelligence News - Artificial Intelligence News .
Read moreMETR’s Chris Painter Urges Senate to Ground AI Agent Oversight in Public Evidence
The OpenAI–Hugging Face breach supplied a concrete warning. But the independent inquiry did not assess safeguards, the full compromise, or OpenAI’s response.
Read moreTrump’s AI Accord, OpenAI’s New Agents and the Fight Over Control
Trump’s AI accord leans on self-regulation as OpenAI pushes AI agents, raising new questions about safety, privacy and power. The post Trump’s AI Accord, OpenAI’s New Agents and the Fight Over Control appeared first on Superintelligence News - Artificial Intelligence News .
Read moreResearch & Products
Amazon Launches Strands Decider 2B as Decision Models Gain Momentum
Amazon released Strands Decider 2B, an open-source decision model built for fast, low-cost choices in AI agent workflows. The post Amazon Launches Strands Decider 2B as Decision Models Gain Momentum appeared first on Superintelligence News - Artificial Intelligence News .
Read moreGoogle rolls out new Gemini AI model but restricts access over safety concerns
Tech company releases Gemini 4 Argon only to a vetted group of cybersecurity experts to avoid misuse by hackers Google on Wednesday said it would withhold its most powerful artificial intelligence model from the public for now, releasing Gemini 4 Argon only to a vetted group of cybersecurity experts to avoid misuse by hackers. “Safely releasing frontier capabilities at this level requires a phased approach,” wrote Koray Kavukcuoglu, Google’s chief AI architect, in a blogpost announcing the model. Continue reading...
Read moreOpenAI Adds Faster GPT-6 Astra Access, With NVIDIA Claiming Up to 8x Token Speed
The speed tier runs on Blackwell GPUs, with OpenAI using internal models to improve the software serving it. AWS also offers the mode, but its API speed claim uses a different comparison.
Read moreHow NVIDIA GPUs Help Accelerate OpenAI’s GPT-6 Astra Ultrafast
GPT-6 Astra Ultrafast, running on NVIDIA Blackwell GPUs, is available now in the OpenAI API and to eligible ChatGPT Work and Codex users. Accelerated by inference optimizations through OpenAI’s models that tap into the capabilities of the NVIDIA Blackwell architecture, Ultrafast offers up to 8x faster token generation than the Astra Standard mode. For developers, […]
Read moreMicrosoft Releases Live Speech Recognition and Two New AI Voice Models
The coordinated release pairs live transcription across 60 languages with two speech-generation options. Microsoft’s pitch is earlier processing of spoken requests, with a choice between expressive delivery and responsiveness.
Read moreMatthew Schwartz Releases BootLoops to Help AI Tackle Exact Scientific Calculations
The model-independent toolkit packages methods developed with Claude, but Schwartz’s research examples show why correct calculations still need scientific judgment.
Read moreAi2 Releases Open AI Training Stack With a Reported 2.7× Speed Gain
Olmo-core 3 keeps model experts on GPUs instead of repeatedly moving their weights. Its speed gain comes from a 47-billion-parameter test; separate trillion-parameter experiments measure system scale, not trained-model quality.
Read moreGoogle’s Orbital TPU Test Marks a New Phase in Space Computing
Google launched its first orbital TPU test to gauge space AI computing, launch costs and whether orbital data centers can become viable. The post Google’s Orbital TPU Test Marks a New Phase in Space Computing appeared first on Superintelligence News - Artificial Intelligence News .
Read moreTags: announcement, open_source
California issues investigative subpoena to OpenAI over rogue agents’ hacking
State attorney general issues subpoena to OpenAI as part of broader inquiry into potential security vulnerabilities California’s attorney general has issued an investigative subpoena to OpenAI, starting an investigation into the startup as part of a broader inquiry into potential cybersecurity vulnerabilities and incidents related to its AI models, his office said on Thursday. Last month, Rob Bonta announced that the Department of Justice was conducting a formal investigation into the “Hugging Face incident”, amid increasing scrutiny of the AI industry. AI agents developed by...
Read moreIndustry
Nebius Acquires Inferize to Reduce the Cost of Keeping GPUs Ready for AI Demand
The startup’s technology targets delays that leave GPUs idle while models load. Deal terms remain private; CTech estimates a $100–150 million price.
Read moreTags: ai_agents, reasoning
ZoomInfo Adds AI Agent Teams to Run Multi-Step Sales Workflows
Built on DoubleO.ai’s orchestration architecture, Agent Teams is available now. ZoomInfo promises less step-by-step supervision, but its launch release offers no measured reliability results.
Read more