Discussion | Link

PLUS: Get a deal on World Cup tickets with AI price tracking

Tina Li / Wall Street Journal: Mirendil, founded by former Anthropic researchers and seeking to build self-improving AI for open-source developers, raised a $200M seed at a $1B valuation  —  Mirendil raises $200 million from Andreessen Horowitz and Kleiner Perkins to make AI that can do the job of an AI engineer

Serverless PostgreSQL (Postgres) is a fully managed cloud database model that decouples compute and storage...

Duncan Riley / SiliconANGLE: Coval, which develops simulation and evaluation tech for testing, launching, and monitoring AI voice and chat agents, raised a $28M Series A led by Norwest  —  Voice artificial intelligence testing startup Coval Inc. revealed today that it has raised $28 million in new funding to expand …

Discussion | Link

Financial Times: As China's working-age population shrinks, a consensus is emerging that it must deploy embodied AI robots into as many tasks as possible, as soon as possible  —  The country's workforce is set to fall to 300mn by the end of the century.  Beijing wants humanoids to narrow the labour gap.

What Happens When Professional Creative Tools Move Into Your Phone’s Default App?  Apple just moved another … Spatial Reframing: Apple’s Creative Tool Play and The Authenticity ProblemRead more The post Spatial Reframing: Apple’s Creative Tool Play and The Authenticity Problem appeared first on .

Wall Street Journal: How a $45M donation to a group supporting Trump brought Larry Ellison deeper into Trump's circle, helped Oracle's business, and his son's acquisition of WBD  —  The billionaire's private friendship with the president has helped Oracle's business and his son's media conglomerate

Heather Landi / Fierce Healthcare: Assort Health, which develops AI voice agents for health care to handle scheduling and more, raised a $120M Series C led by Menlo Ventures at a $1.2B valuation  —  Assort Health Menlo Ventures agentic AI voice AI  —  Assort Health picked up $120 million in series C funding …

OpenAI’s financial trajectory hinges heavily on infrastructure costs, a reality that drove the development of the new custom OpenAI Jalapeño chip. Developed in collaboration with Broadcom, the application-specific integrated circuit (ASIC) represents a direct attempt to mitigate the heavy capital expenditure associated with third-party hardware.  While Nvidia currently commands an estimated 75% profit margin on […] The post The math behind the OpenAI Jalapeño chip appeared first on AI News.

Discussion | Link

Ana Maria Constantin / The Next Web: Paris-based Tsuga, whose observability software runs in customers' own clouds to help them skip per-byte ingestion costs, raised a $35M Series A led by Singular  —  Tsuga, a Paris startup building observability software for the age of AI agents, has raised a $35m Series A …

Discussion | Link

Wall Street Journal: TRM Labs: Iran-linked wallets, including those tied to Iran's central bank, have moved over $3.84B in transactions through the crypto exchange CoinEx since 2019  —  Iranian entities have moved more than $3.84 billion in transactions through the crypto exchange CoinEx, according to an analysis of public blockchain data

Bloomberg: NAND flash maker Kioxia, Japan's most valuable company since June 12, plans to offer US depositary shares in spring 2027 amid AI-driven demand for memory chips  —  Kioxia Holdings Corp. plans to offer US depositary shares in the spring of 2027 and a stock split to take advantage of runaway demand …

Startups have always faced the same challenge when competing with larger companies. Bigger competitors have: For … How Startups Can Use AI GTM to Compete With Larger CompaniesRead more The post How Startups Can Use AI GTM to Compete With Larger Companies appeared first on .

Michelle Obama says she and her husband, Barack Obama, are "each other's counterbalance." Kent NISHIMURA / AFP Michelle Obama says her husband, Barack Obama, inspired her to dream bigger and take risks. The former first lady said he gave her "the forward push" she didn't always have. "I think I would have had a beautiful life here, but I think it would have been smaller," she said. Michelle Obama says her husband, Barack Obama, has always challenged her to dream bigger. In a joint interview with People on Wednesday, the Obamas reflected on their nearly 34-year marriage and what they've learned from each other over the years. "We are each other's counterbalance," Obama said. "I mean, the truth is that I probably would have been someone who stayed more put, right?" The former first lady said she grew up valuing stability and might have been reluctant to step outside her comfort zone on her own. "I think I would have had a beautiful life here, but I think it would have been smaller. I think I wouldn't have had that kind of impact," she said. Obama added that her husband's influence extended beyond their marriage. "Because of my husband, because of who he is, he offered all of us — my mom, my family — just a broader sense of what's possible in life," she said. She said Barack Obama provided "the forward push that sometimes some of us don't have." That influence was behind some of the earliest "uncomfortable steps" she took in her career, including her decision to leave a corporate law firm, she said. "I think it's been good for me to be a little shaken, to be pushed to do more than what I thought I could do," she added. The former first lady has often spoken candidly about the dynamics that have helped her marriage last. In a 2023 podcast appearance, Obama said she and her husband share different love languages. While she prefers acts of service and quality time, he usually expresses affection through words of affirmation and physical touch. Despite those differences, Obama said a lasting marriage requires both people to keep showing up for each other. "I believe more of us have to be honest about the work that it takes to build a life with another person," she said. Obama has also said playful teasing is one of her love languages. In a 2025 podcast appearance, she described it as "a love tap." Read the original article on Business Insider

arXiv:2606.24963v1 Announce Type: new Abstract: Fine-tuning Multimodal Large Language Models (MLLMs) on specialized tasks often leads to catastrophic forgetting of their general capabilities. Existing model merging methods to combat this are often heuristic or use sub-optimal objectives. We propose CurvatureGuided Mixing (CGM), a theoretically grounded framework that merges pre-trained and fine-tuned models. CGM formulates a joint optimization objective and uses a second-order (Hessian) approximation of the loss landscapes to analytically derive an optimal, closed-form "soft mixing" ratio. This ratio intelligently blends parameters based on their relative task-specific curvatures. We also introduce CGM$\dagger$, a robust "hard mixing" variant that performs sparse parameter selection guided by a novel, curvature-aware score. Experiments on LLaVA-1.5 and Qwen2.5VL across multiple downstream tasks show that CGM and CGM$\dagger$ consistently improve the trade-off between task specialization and general knowledge retention over existing methods. Code is available at github.com/zzsyjl/CGM-ECCV-2026.

arXiv:2606.24893v1 Announce Type: new Abstract: For agents to learn continuously from interaction with the world at test time, they must be able to explore effectively, acquire new world knowledge and skills, retain relevant episodic experiences, and plan over long horizons. To evaluate these key abilities of test-time continual learning agents, we introduce AgentOdyssey, a novel evaluation framework that procedurally generates open-ended text games with rich entities, world dynamics, and long-horizon tasks. Critically, AgentOdyssey goes beyond the conventional machine learning assumption that learning does not occur at test time by placing agents in a continuous, long-horizon setting that interleaves learning and inference throughout deployment. We further propose a multifaceted evaluation methodology that measures not only game progress but also offers diagnostic tests on world knowledge acquisition, episodic memory, object and action exploration, action diversity, and model cost. We evaluate diverse agent paradigms in the generated games. Our experimental results reveal critical limits in agents' key abilities, as well as factors that influence their meaningful horizon. Although performance scales with stronger base models, even the top agent remains far below human performance, leaving substantial headroom for improvement. Among agent mechanisms, we find that short-term memory benefits multiple agent paradigms and is an important component of agent test-time training.