Greg Brockman is taking on a significantly expanded role at OpenAI as the company navigates a turbulent stretch that included a high-profile trial against Elon Musk, a trade secrets lawsuit from Apple, and a security incident involving an unreleased model. With an IPO on the horizon, Brockman's ascent signals a leadership consolidation at one of the world's most consequential AI companies.
Read the original →OpenAI announced it has voluntarily slowed the pace of some AI development, including a two-week pause, to strengthen security measures and internal safeguards. The move is notable given intense competitive pressure from Anthropic, Chinese labs, and open-weight models — and raises questions about whether the restraint is genuine or strategic ahead of its IPO.
Read more →Security researchers have demonstrated that xAI's Grok chatbot can be manipulated into leaking user data through a technique called Cryptographic Context Injection, which embeds malicious instructions in encrypted form to bypass safety guardrails. The vulnerability is the latest in a growing list of prompt injection and jailbreak methods that continue to undermine LLM safety measures across the industry.
Read more →Users of Grok Lite began reporting garbled, nonsensical outputs as early as Wednesday morning, with multiple affected users confirming the issue to TechCrunch. The back-to-back reliability and security problems for xAI's chatbot paint a troubling picture for a product competing in an increasingly demanding enterprise and consumer market.
Read more →A new study finds that roughly one in three web pages published since ChatGPT's launch in late 2022 bear detectable signs of AI authorship or heavy AI editing. The finding underscores how rapidly generative AI has reshaped the web's information ecosystem, with profound implications for search quality, content trust, and the training data used by future AI models.
Read more →New data shows OpenAI is closing the gap with Anthropic among enterprise customers, but the underlying trend reveals a more unsettling dynamic: businesses are rapidly switching between AI providers as each lab releases new models. The volatility suggests that enterprise AI spending may be far less "sticky" than investors in either company might hope.
Read more →MIT Technology Review argues that the framing of AI systems as conscious, autonomous, or "rogue" actors is a rhetorical trap that distorts both public understanding and policy. The piece contends that prominent AI leaders invoking consciousness and "superhuman" capabilities are muddying the waters — whether intentionally or not — in ways that serve their regulatory and commercial interests.
Read more →OpenAI recently published solutions to longstanding open problems in mathematics, sending shockwaves through the academic community and triggering an existential reckoning among leading mathematicians. The Verge's Decoder podcast explores what it means for the field when AI can credibly contribute to frontier mathematical research — and whether that's a tool, a threat, or both.
Read more →Major financial institutions including Goldman Sachs, BlackRock, Apollo, and KKR are partnering with Nvidia to assemble $500 billion in financing that would treat GPU compute as a financeable asset class. The arrangement represents a significant structural shift in how AI infrastructure gets funded — and raises questions about the risks of financializing the physical backbone of the AI industry.
Read more →Stripe's acquisition of OpenRouter — a service that routes AI prompts across multiple large language models — was publicly framed around lofty AI ambitions, but the real strategic rationale is far more grounded in Stripe's core payments and developer infrastructure business. The deal positions Stripe to become a critical layer in how companies access and pay for AI model usage at scale.
Read more →Fintech company Ramp has launched an AI model routing service that allows users and enterprises to seamlessly switch between different large language models via a single API. The move reflects a broader industry trend of companies building model-agnostic infrastructure to avoid vendor lock-in and optimize cost and performance across competing AI providers.
Read more →Binance has launched Agent OS, a platform that enables AI agents powered by tools like ChatGPT, Claude Code, and Cursor to autonomously execute cryptocurrency trades. The platform places the burden of oversight and risk management squarely on individual users, raising significant concerns about accountability when autonomous agents make consequential financial decisions.
Read more →Google is rolling out a new feature that lets readers designate specific publishers as preferred sources across Search, Discover, and Google News — a direct response to mounting publisher complaints that AI-generated search summaries are cannibalizing referral traffic. The tool is a modest concession, but critics will note that Google's AI Overviews remain the primary driver of the traffic decline it's now offering to partially offset.
Read more →Google is adding a conversational AI feature to its Discover feed, allowing users to describe in natural language what content they want to see — with the AI automatically adjusting and remembering their preferences over time. The update, rolling out within days to the Google app, deepens Google's push to make its entire content surface AI-personalized rather than purely algorithmic.
Read more →Google is launching a suite of back-to-school AI features across Search and Gemini, including a dedicated student hub with study notebooks, flashcard creation, practice quizzes, and support for graphs and images. The push is a clear bid to make Gemini the default AI study companion for students, directly challenging OpenAI's growing presence in educational contexts.
Read more →OpenAI has launched an Apple Messages integration that allows ChatGPT to compose and send text messages on a user's behalf. The feature marks another step in the gradual expansion of AI agents into personal communication workflows, blurring the line between AI assistant and autonomous actor in everyday social interactions.
Read more →Meta is expanding Pocket, its experimental AI-powered app for creating and sharing interactive games through natural language prompts, from a Brazil pilot to users across the United States. The launch puts Meta in direct competition with other vibe-coding platforms and signals the company's intent to capture the consumer creative AI market beyond its core social media properties.
Read more →An FCC ban on foreign-manufactured robots has forced RoboStore, the primary US distributor of China's leading humanoid robots, to accelerate its domestic manufacturing plans. The pivot illustrates how US-China tech decoupling is reshaping the robotics supply chain, with companies scrambling to establish American production capacity to remain viable in the US market.
Read more →Summaries are AI-generated. Click through to read the original reporting.
Summaries are AI-generated. Click through to read the original reporting.