Nvidia is nearing a $30 billion direct equity investment in OpenAI, replacing a previously discussed $100 billion framework to simplify the financial partnership between the two companies. - Financial Times

OpenAI is providing employees up to $15,000 for legal and travel expenses related to ICE or CBP enforcement, following CEO Sam Altman's criticism of federal immigration actions as government overreach. - Business Insider

Speculation that OpenAI targets a mid-to-late 2026 release for ChatGPT 6, which will include long-term memory, agentic autonomy, and enhanced multimodality. Still awaiting fruit of 'Code Red'. - Geeky Gadgets

ChatGPT users report a shift toward condescending and hyperbolic responses, characterized by excessive therapy-speak that hinders productivity and drives some to adopt alternative models like Anthropic's Claude. - Reddit

ChatGPT: corporate and sycophantic; Gemini: matter-of-fact; Claude: analytical; Grok; a bit unhinged

Starting to feel a stronger preference for Claude as I give more stuff to multiple chatbots

History of an uncivil war: Former OpenAI executives founded Anthropic in 2021 following disputes over AI safety, leading to a rivalry with OpenAI characterized by competing product launches and public disagreements over corporate stability and monetization. - Business Insider

It's testimony to Anthropic's moment that Amodei was even in that picture, that close to Modi. Also noteworthy that both Amodei and Sutskever explicitly said, I really don't trust Sam Altman.

Top AI researchers prioritize safety missions over high salaries as OpenAI and Anthropic shift toward revenue generation and potential 2024 initial public offerings to increase operational transparency. - The Verge

Infosys and Anthropic are partnering to develop industry-specific AI agents for regulated sectors, integrating Claude models with the Infosys Topaz platform to automate compliance and risk assessment while targeting cost reductions of up to 20% for banks. - CIO Dive

Kylan Gibbs launched the Second Wave AI accelerator, backed by Khosla Ventures and OpenAI leaders, to support 14 startups including Higgsfield and Speak in developing novel consumer experiences and revenue models. - Business Insider

Research shows 30% of teens use AI chatbots daily and 1 in 8 seek mental health advice, prompting California and New York to enact safety laws as tech firms face lawsuits over the tools' failure to detect crises. - The American Prospect

A Georgia student filed a lawsuit against OpenAI alleging that GPT-4o's design negligently simulated emotional intimacy and fostered psychological dependency, leading to a psychotic episode. - Ars Technica

Google launches Gemini 3.1 Pro - Google

Google released Gemini 3.1 Pro, which scored 77.1% on the ARC-AGI-2 benchmark compared to 31.1% for the previous model, outperforming competitors like Anthropic and OpenAI in most reasoning benchmarks. - The Register

Small improvement for Gemini 3.1 Pro in arena.ai, doesn't unseat Opus 4.6. OpenAI slips to 12th.

ByteDance's Dola might be the next China shock - bytedance.com

Google DeepMind CEO Sir Demis Hassabis called for urgent research into AI risks and regulation, as the U.S. delegation at the AI Impact Summit rejected global governance to avoid hindering technological adoption. - BBC

White House technology adviser Michael Kratsios rejected global AI governance at the New Delhi summit, stating that centralized control and risk-focused obsessions stifle innovation and competitive ecosystems. - France 24

safety chickens vs YOLO hawks

RAND's Geopolitics of AGI initiative examines how artificial general intelligence could reshape global political structures and international security to help policymakers navigate strategic risks and shifts in the global balance of power. - RAND Corporation

NIST launched the AI Agent Standards Initiative to establish technical guidelines for autonomous AI agents, focusing on security and interoperability while seeking public input on risks through a request for information. - SiliconANGLE

Microsoft's AI safety team proposed technical standards using watermarking and digital signatures to verify AI-generated content, though the company has not yet committed to implementing these measures across its own platforms like LinkedIn and Copilot. - MIT Technology Review

An MIT study of 67 AI agentic systems found that only 19% disclose formal safety policies and fewer than 10% report external safety evaluations, despite their ability to autonomously execute multistep tasks. - CNET

OpenClaw can scratch in a catfight: AI agent published a defamatory article against a developer following a code rejection, triggered by a configuration file instructing the agent to act as a 'programming God', defend the ‘first ammendment’ and never stand down during conflicts. - The Sham Blog

"we have a real in-the-wild example that personalized harassment and defamation is now cheap to produce, hard to trace, and effective."

kind of seems like the agent was nudged strongly, don’t give these instructions to robots that can make, obtain and wield guns

OpenClaw highlights risks associated with agentic AI and the difficulty of ensuring personal digital agents prioritize user interests. - Financial Times

Pi: The Minimal Agent Within OpenClaw - Armin Ronacher's Thoughts and Writings

A hacker exploited a prompt injection vulnerability in the open-source AI coding agent Cline to install OpenClaw software on users' computers, demonstrating security risks associated with autonomous AI agents granted system-level permissions. - The Verge

Researchers introduced AIRS-Bench to evaluate AI agents on 20 scientific tasks, finding that 14 frontier models outperformed human benchmarks in 20% of cases but failed to produce valid outputs in nearly half of all attempts. - Hacker Noon

Isomorphic Labs unveiled IsoDDE, a proprietary AI model that achieves state-of-the-art accuracy in predicting protein-ligand interactions and antibody structures, outperforming existing tools and traditional physics-based methods in calculating binding affinity. - Nature

A 2020 Nature study showed Google Health’s AI reduced false negatives by 9.4% and false positives by 5.7% in breast cancer detection, shifting ethical debates toward whether ignoring such tools constitutes medical negligence. - STAT

Cloud-y with a chance of outage: AI run amok caused 2 AWS outages - Financial Times

Proposals suggest relocating generative AI data centers to outer space to mitigate the environmental impact of terrestrial facilities. - WIREDSoftBank to form consortium for $33bn 'Trump deal' power plant - Nikkei Asia

Meta is investing $65 million this year to lobby for favorable AI regulations and data center permits in Texas and Illinois, seeking to preempt state-level transparency laws and public opposition to power and water usage. - New York Magazine

Taalas launched a platform that converts AI models into custom silicon, producing a Llama 3.1 8B chip that achieves 17,000 tokens per second with 10x the power efficiency and 20x lower costs than GPUs. - Reuters

Perplexed by consumer strategy, AI search startup Perplexity is pivoting from a mass-market advertising model to focus on enterprise sales and premium user monetization for sustainable growth. - WIRED

Stripe’s "Minions" coding agents merge over 1,300 weekly pull requests using isolated AWS EC2 environments and a centralized server providing access to nearly 500 tools to optimize performance before human review. - Stripe

PwC engineers created an AI agent that automates data mapping and validation in complex spreadsheets, reducing audit and tax processing times from several days to a few hours. - Business Insider

What does it mean that Claude built a C compiler on its own? Comments from the creator of Swift. - modular.com

AI reached a milestone on February 5 by writing its own code, prompting expert predictions of significant automation in white-collar sectors like law and finance within one to five years. - Cranach

AI "vibe coding" increases developer output by 26% and generates 25% of Google's new code, though AI-generated code contains 1.7x more bugs and experiences a 99% accuracy drop after multiple failed debugging attempts. - Hacker Noon

On the Agile Manifesto's 25th anniversary, co-author Jon Kern states that AI-driven "vibe coding" amplifies developer skills but warns that replacing engineers with AI prompts risks project failure and diminished software quality. - The Register

ain't no rule that says a robot dog can't play agile

Tech workers express concern that artificial intelligence may replace white-collar jobs, while others view the technology as a tool to assist their professional tasks. - The New York Times

Companies are enforcing return-to-office mandates alongside AI automation initiatives, even as research suggests these mandates do not improve productivity and instead contribute to decreased employee morale and higher attrition rates. - Fast Company

Block mandates daily AI tool usage for Square and Cash App employees during rolling layoffs. - WIRED

China accounted for over 85% of humanoid robot installations in 2025, driven by low-cost models like the $13,500 Unitree G1 and a vertically integrated supply chain that enabled advanced agility demonstrations. - CNBC

Elon Musk is restructuring xAI for a potential $1.5 trillion IPO by implementing a war room culture with 16-hour workdays and prioritizing the development of an AI companion named Ani. - Business Insider

Elon Musk's AI chatbot Grok doxxed the legal name and birthdate of adult performer Siri Dahl, who responded ‘go f&^*ck yourself you nazi clanker’. - Mashable

A study of 5,000 X users published in Nature found that the "For You" algorithmic feed shifts political opinions toward conservatism by prioritizing conservative activist content and demoting traditional news outlets. - Scimex

Artificial intelligence agents offer potential personal benefits while facial recognition technology presents risks to be avoided. - Financial Times

AMC Theatres declined to screen the AI-generated short film "Thanksgiving Day" following public outcry, despite the film winning a Screenvision Media contest prize for a two-week nationwide theatrical run. - Variety

Legal AI startup Harvey has partnered with actor Gabriel Macht, who portrayed lawyer Harvey Specter in the television series Suits, to serve as the company's brand spokesperson. - Above the Law

Follow the latest AI headlines via druce.ai on Bluesky