AI Safety Concerns, Big Tech's Code of Conduct, Agentic AI Growth
Microsoft issues a 'humanist AI code of conduct' amid growing safety concerns from OpenAI and Anthropic CEOs.
Sunday, September 13, 2026
This week's stories caught my attention, especially the debate about AI risks to humanity and the growing pressure on tech companies to prioritize human well-being and control in AI systems. I'm also interested in the calls for a slowdown in AI development and the responses from companies and governments. It's an exciting and complex time for the AI industry.
— Razi
🔥 Top story
Microsoft Publishes 37-Page 'Humanist AI Code of Conduct'
Microsoft has released a comprehensive 37-page 'humanist AI code of conduct' in response to escalating AI safety concerns. This move comes after Anthropic CEO Dario Amodei called for a coordinated slowdown in AI development, following warnings that AI progress might outpace our ability to ensure safe deployment. The document outlines principles and practices aimed at prioritizing human well-being and control in AI systems. This initiative highlights the growing pressure on major tech companies to address ethical implications as AI models become increasingly sophisticated and pervasive.
OpenAI and Anthropic CEOs Warn of AI Risks to Humanity
CEOs of leading AI companies, OpenAI and Anthropic, have issued new warnings about the potential risks of advanced AI to humanity, reigniting a long-standing debate. These statements follow recent reports from researchers suggesting that the rapid advancement of AI models could soon exceed current safety protocols and human oversight capabilities. The renewed focus from these prominent figures underscores a critical period for the AI industry as it grapples with balancing innovation and responsible development. This debate is likely to influence future regulatory discussions and public perception of AI.
Temporal Raises $550M at $12.55B Valuation for AI Agent Resilience Software
Temporal, an open-source software provider enabling applications and AI agents to recover from failures, has successfully raised $550 million in a late-stage funding round. Led by Lightspeed, this investment more than doubles the company's valuation to $12.55 billion. Temporal's technology is critical for building robust and reliable AI systems, allowing them to maintain state and resume operations seamlessly after interruptions. This significant funding highlights the growing market demand for infrastructure that ensures the resilience and fault tolerance of complex AI applications, especially as agentic AI systems become more prevalent.
OpenAI Missives Drive Debate on AI Cyber Defense Responsibility
Recent communications from OpenAI have spurred a debate regarding who should bear the primary responsibility for AI cyber defenses. As AI models become more powerful and integrated into critical infrastructure, the potential for sophisticated cyber threats escalates. This discussion involves governments, AI developers, and end-users, each with a role in securing these complex systems. The ongoing dialogue highlights the urgent need for clear guidelines and collaborative strategies to protect against AI-enabled cyberattacks and ensure the integrity of AI deployments.
Copado Extends Agentia AI DevOps for Salesforce with Headless Automation
Copado Inc., a low-code DevOps solution provider for Salesforce, has expanded its Agentia platform by integrating Headless automation. This enhancement brings Agentia, Copado's AI-powered AgentOps platform, directly into developer tools and operational workflows, streamlining the planning, building, testing, and releasing of software across the Salesforce ecosystem. The headless capability allows for greater flexibility and direct integration with existing CI/CD pipelines and development environments. This development is crucial for enterprises leveraging Salesforce, enabling more efficient and automated AI-driven software delivery.
AI Leaders Call for Slowdown, Trump Team Cites Company Responsibility
Following calls for restraint from AI leaders like Sam Altman and Elon Musk, and Dario Amodei's plea to Washington to slow AI development, Donald Trump's team has stated that the responsibility for managing AI risks lies with the companies themselves. This political response underscores a divergence in approaches to AI governance and regulation. While some advocate for industry-wide pauses and governmental intervention, others emphasize self-regulation and maintaining a competitive edge, particularly against countries like China. This ongoing tension will shape future policy decisions and the pace of AI innovation.
Topic in, narrated explainer video out. A Claude Code / Codex skill that turns any topic into a black-canvas motion-graphics explainer video with TTS voiceover, subtitles and a chapter progress bar…
Free open-source project designed for turning youtube-viedos into viral short videos. Highlight detection, subtitles, translation, voiceover, all in one for your content.
The paper identifies a fundamental tension when scaling RL for automatic research agents, where environment execution dominates the training cost and becomes the bottleneck. This matters to people building with AI because it prevents them from scaling their agents to larger sizes. The paper proposes World Model RL, which replaces environment execution with a world model to remove this bottleneck, but it leaves open the question of how WMRL performs in more complex or dynamic environments.
arXiv · 452 upvotes on Hugging Face Papers · Sep 14
UAE Pass is OAuth 2.0 until the token call, which takes its parameters in the query string and returns no id_token. The staging traps, a working C# client, what the SOP levels actually mean, why linking must key on UUID — and the design and copy rules that decide how many rounds of assessment you go through.
🧠 Fun Fact: A recent study found that the energy consumed by training a single large AI model can be equivalent to the lifetime carbon emissions of five cars.
