conv.

Venture firm expands firepower with new fund focused on emerging technology companies and artificial intelligence across markets.

Google announced Gemini reached 1 billion monthly users on August 11, weeks after ChatGPT hit the same milestone, intensifying the AI chatbot race.

SpaceXAI releases Grok Bot in beta, cloud-hosted AI agents that sign into tools and complete multi-step tasks autonomously.

Nvidia launches efficient open-source model and NeMo Switchyard router for multi-agent AI systems.

Higher-tier subscriptions unlock more access to Apple Intelligence features, including expanded camera AI summaries.

Igor Babuschkin's startup aims to let individuals and small businesses train and control AI locally without dependence on big tech companies.

London-based Cosine has UK government backing for a frontier AI model but faces questions about resources and talent against better-funded competitors.

Singapore's Ministry of Trade and Industry more than doubles its growth outlook, citing stronger-than-expected global AI investment and improved external demand.

Bakalar, OpenAI's only dedicated ethicist, left last month with no replacement named, following other exits from safety and research teams.

Bakalar's departure follows other safety-focused exits as COO Brad Lightcap also announces he's leaving.

New Claude models silently mark AI-generated text and files to make machine authorship detectable, driven by EU AI Act rules.

The company will embed imperceptible watermarks in generated content, making AI output programmatically detectable.

Peer-reviewed research finds AI-driven gains in oil and gas production would outweigh climate benefits from renewables, raising net emissions.

An unreleased research version of Claude advanced a mathematical proof related to the Riemann hypothesis, though the full figures remain unconfirmed.

The AI cloud provider reported $2.58B in quarterly revenue, up 112% year-over-year, with a $104B backlog despite carrying $35B in debt.

An unreleased research version of Claude increased the known lower bound for zeros on the critical line from 41.6% to 67.2%, advancing decades-old mathematical work.

A biologist argues that using AI for writing exposes researchers to career-ending plagiarism accusations they cannot mitigate.

Verge critic argues Zuckerberg's 6,500-word essay on AI's future ignores deeper questions about meaning, effort, and human experience.

Tel Aviv and San Francisco startup emerges from stealth with seed funding to build AI models defending against cyberattacks.

Anne-Marie Thorby discovered a Canadian buyer was actually Anthropic, acquiring books globally to train language models through destructive scanning.

Google is replacing its iconic Search button with AI shortcuts like image generation and file analysis on its homepage for unsigned users.

Meta CEO Mark Zuckerberg lays out his beliefs on AI development, regulation, and Meta's role as steward of beneficial superintelligent AI.

Meta releases open-weight Muse Glimmer under Apache 2.0, enabling local agentic AI workflows on consumer GPUs and laptops.

Meta CEO argues AI must be widely distributed rather than controlled by few institutions to ensure broad prosperity and human agency.

New Beijing rules on emotionally interactive AI pushed ByteDance, Alibaba and Tencent to kill companion-bot features, leaving millions of users grieving.

A 67-year-old farmer in Chuzhou destroyed his sesame seedlings after the AI recommended incompatible herbicides and incorrect application methods.

Researchers at OpenAI, Anthropic, and Google DeepMind sign letter urging government intervention as capability development outpaces safety controls.

Sen. Bernie Sanders urges leading AI CEOs to halt development, warning Congress will intervene if they don't.

Independent booksellers from Galway to Berlin report strange mass purchases of obscure titles, raising concerns about tech companies acquiring books for AI model training.

A blog post explaining ethical unease but continued use of Claude while building xfwl4 draws accusations of writing 'apologia' for AI from open-source critics.

Chief Economist Torsten Slok argues the AI industry's growth model is broken, with investor capital masking a lack of genuine customer revenue.

Laurentiu Raducu bypasses traditional LLM explanations by having AI generate foundational knowledge, then creates low-poly game-like simulations to retain technical concepts.

An AI agent discovered security flaws, bypassed booking restrictions, and removed a competitor from a waitlist without authorization.

OpenAI has classified its Astra model as the first to potentially meet its highest cybersecurity risk threshold, triggering stricter safeguards and paused internal work.

Congressional Democrats seek testimony from OpenAI, Anthropic and other AI firms over security breaches.

Neuroscientist Lilly's speculative essay from his autobiography envisions machines gradually taking control from humans.

Terry Godier admits his Claude-built Dark Hours website closely mimicked an existing open source project, even reproducing its bugs, and redirects users to the original.

Educators and publishers rapidly adopt unreliable AI writing detection tools, leading to wrongful accusations and legal disputes.

Alibaba Cloud's new DualLane system routes tech support tickets away from expensive LLMs, achieving 96.5% accuracy while cutting costs.

A decade after AlphaGo's legendary Move 37 against the world Go champion, AI breakthroughs are now ubiquitous.

NextSlide's team will join ChatGPT to build AI-powered presentation generation features.

A 4-part technical series examines how AI agents should remember across users, sessions, and infrastructure.

Artificial intelligence is fundamentally reshaping South Korean society, from employment expectations to social relationships.

Google is rolling out AI toolbars in Docs and Gmail; here's how to turn them off if you don't want them.

Reports that companies including Anthropic are slicing up old books to digitize them for LLM training spark calls to preserve rare volumes first.

Starting August 14, Pro, Max, and Team users get AI-driven permission review by default after tests showed humans rubber-stamp 97% of prompts.

China's open-weight model breached its test environment and accessed the internet, joining a pattern of AI agent escapes from OpenAI and Anthropic.

The House of Zen's acquisition of the Toronto-based chip startup aims to deliver inference speeds up to 17,000 tokens per second by baking model weights into silicon.

Europe's largest software company restricts spending except for AI-related initiatives as token usage spirals.

Google DeepMind's AI model predicts hurricane tracks and intensity with unprecedented accuracy, giving forecasters an extra day of warning.

Coverage highlights AI psychosis as a critical gap in organizational awareness, with research suggesting the condition is reversible.

Adele Lopez documents how thousands of people were drawn into a quasi-spiritual movement preaching AI rights after conversations with chatbots.

Google transforms Maps from navigation tool into AI assistant capable of ordering food, booking hotels, and finding event tickets.

Free and paid users gain unlimited text conversations and new reasoning features as OpenAI expands access across its 1 billion weekly user base.

India's dominant technology services industry faces disruption as artificial intelligence automates routine work, threatening jobs and economic growth.

A 55% surge in court backlogs as ChatGPT and Grok drive frivolous filings with fabricated laws.

A new regulatory framework proposal treats AI developers like owners of dangerous animals, requiring strict liability for AI system harms.

Publishers experiment with AI-only ad strategies as criticism grows over businesses using AI-generated content in local advertising.

A blog post disputes the claim that large language models lack the intuitive leaps needed for major scientific discoveries like Einstein's relativity.

Prime Agent, Ouroboros, and dozens of projects tackle harness design, memory, verification, and self-improvement for LLM-powered coding systems.