Story cards stay below the sticky dock while audio, chapters, date, and brief navigation remain accessible.
Salesforce announced it will acquire Fin, the AI agent platform resolving customer queries across live chat, WhatsApp, SMS, phone, and Slack, for $3.6 billion. The deal removes one of the fastest-growing independent agentic customer service challengers from the market and cements Salesforce's dominance in enterprise AI CX. For builders in this space, the independent competitor pool just lost a very significant name.
Meta rolled out "AI Mode" on Facebook today, a search experience that uses Meta AI to synthesize answers from public posts, Groups, and Reels across its platforms — no links, just a conversational response grounded in what people are actually discussing. The feature draws from Facebook, Instagram, Threads, and other Meta properties, making social data the backbone of the company's AI search offering. It's Meta's most aggressive move yet to differentiate its AI on the one asset rivals can't replicate: its social graph.
U.S. District Judge Rita Lin in San Francisco dismissed Elon Musk's xAI trade secret lawsuit against OpenAI with prejudice Monday, closing the case for good — xAI cannot refile. The suit alleged that former xAI engineer Xuechen Li disclosed Grok source code to OpenAI during recruitment; Judge Lin said xAI simply failed to prove it, calling further attempts "futile." The ruling delivers a decisive legal defeat to Musk in his ongoing rivalry with Sam Altman and ends one of the more dramatic side plots in the AI wars.
OpenRouter's Fusion API dispatches prompts to multiple AI models in parallel, then uses a judge model to synthesize the outputs — and its benchmark results are turning heads. On the DRACO research benchmark, a budget panel of Gemini 3 Flash, Kimi K2.6, and DeepSeek V4 Pro scored 64.7%, beating GPT-5.5 at 60.0% and Claude Opus 4.8 at 58.8% individually. The result challenges the "just buy the most powerful model" strategy and suggests that model diversity may matter as much as raw scale for complex research tasks.
Indian AI startup Sarvam raised $234 million at a $1.5 billion valuation, with $150 million led by HCLTech, making it India's newest AI unicorn. The timing is pointed: the funding comes days after the Fable disruption exposed how dependent Indian enterprises — including TCS with 50,000 affected employees — are on foreign AI infrastructure. Sarvam focuses on multilingual AI built specifically for Indian languages and enterprise workflows, precisely the sovereign capability India has been urgently seeking.
With Altman, Amodei, and Hassabis in the room in Évian-les-Bains, OpenAI's chief global affairs officer Chris Lehane said today the company expects participants to leave the summit with an agreed package of voluntary commitments. Youth safety tops Altman's personal agenda; cyber and biological frontier AI risks are also on the formal agenda, with a working lunch Wednesday bringing world leaders and tech executives together. Whether voluntary commitments without enforcement mechanisms amount to anything meaningful is the question hanging over the room.
NewCore emerged from stealth Monday with $66 million in seed funding at a $300 million valuation, led by Cyberstarts with Index Ventures and Evolution Equity participating. The startup's focus: as companies deploy AI agents at scale, there's currently no reliable way to verify which agent did what, on behalf of which user, against which system. NewCore is building the identity and authentication layer that agentic AI deployments don't yet have — a problem that's becoming a production blocker for enterprise rollouts.
Arcade, the startup that authored the MCP authorization spec and runs AI agent actions across the Fortune 500, raised $60 million in Series A funding led by SYN Ventures, with Morgan Stanley and Wipro joining strategically. The company targets the governance gap: security teams can't currently trace which agent took which action on behalf of which user. On a day when NewCore also raised for agent identity, the pattern is unmistakable — the agent governance market is capitalizing fast.
Yale's Digital Ethics Center published a proposal today for a Contextual Copyleft AI License that would treat AI models as derivative works of their training data, requiring developers who train on open-source code to publicly disclose their model architecture and training data in return. The framework faces the classic collective action problem — it only works if enough developers adopt it — and has no current enforcement mechanism. But open-source licensing norms have shaped software development for decades; if this framing gains traction, it could become a real constraint on how labs source training data.
A TechCrunch analysis published Monday finds that 363 layoff events at tech companies in 2026 have affected nearly 150,000 workers, with AI cited as the primary driver across the board. Companies posting record profits are simultaneously reducing headcount, using AI productivity gains as the public explanation, with no coordinated policy response emerging at any level of government. The piece calls it a powder keg — and the day's other nine stories suggest the pressure is only building.
