OpenAI前员工悄然投资新基金，规模或达1亿美元OpenAI alums have been quietly investing from a new, potentially $100M fund

1. Iran has issued a warning of potential retaliatory attacks targeting data centers across the Middle East in response to U.S. threats and military actions. According to a video released by Iranian military spokesperson Ebrahim Zolfaghari, Iran will respond to any U.S. strikes on its civilian infrastructure—such as power and water facilities—with counterattacks on U.S.-linked energy and technology infrastructure in the region. The video specifically highlighted the Stargate data center project in the United Arab Emirates, displaying a zoomed-in image with the message “nothing stays hidden to our sight, though hidden by Google.” Stargate, a $500 billion joint venture between OpenAI, SoftBank, and Oracle announced in January 2025, aims to expand AI data center infrastructure globally but has faced challenges including funding issues and tariff-related costs. The threat follows U.S. President Trump’s ultimatum to strike Iranian civilian infrastructure by the end of Tuesday if Iran does not reopen the Strait of Hormuz, a vital shipping route disrupted since February. Prior missile strikes have already damaged Amazon Web Services (AWS) data centers in Bahrain and an Oracle facility in Dubai. Iran has also named Nvidia and Apple as potential targets. Detailed information limited on the current operational status of affected facilities or specific retaliatory plans.
   Key Takeaways:
   Iran threatens Middle East data centers over U.S. military actions
   Stargate AI project targeted in Iranian military warning video
   Previous missile strikes hit AWS and Oracle data centers
   U.S.-Iran tensions escalate over Strait of Hormuz and infrastructure
   Source: Original Article

1. Robotic machine learning company Generalist has introduced GEN-1, a physical AI system designed to achieve production-level performance across a wide range of manual tasks previously requiring human dexterity. The system demonstrates high success rates in delicate, repetitive operations such as folding boxes, packing phones, and servicing robot vacuums, reaching 99% accuracy at approximately three times the speed of its predecessor, GEN-0. GEN-1 builds on scaling principles similar to those used in large language models, but addresses the unique challenge of limited physical training data by leveraging “data hands”—wearable pincers that record human micro-movements and visual input during manual tasks. Generalist reports collecting over 500,000 hours and petabytes of physical interaction data to train the model. The system adapts to specific robotic platforms in about one hour, enabling rapid deployment. Notably, GEN-1 can recover from disruptions by improvising new actions and synthesizing solutions from diverse experiences, suggesting improved generalization and adaptability in unstructured environments. This advancement signals progress toward autonomous robots capable of handling complex, real-world tasks with minimal reprogramming.

Key Takeaways:
GEN-1 achieves 99% success in delicate mechanical tasks at triple GEN-0’s speed
System trained on 500,000+ hours of human movement data via wearable sensors
Adapts to new robotic platforms in about one hour with minimal fine-tuning
Demonstrates improvisation and problem-solving in response to physical disruptions

Source: Original Article

1. The Trump administration released its proposed 2027 federal budget on Friday, calling for deep cuts across science-focused agencies. The National Science Foundation (NSF) and Environmental Protection Agency (EPA) face the most severe reductions, with budgets slashed by approximately 50%. The National Institutes of Health (NIH), despite oversight by Trump allies, would lose $5 billion from its $47 billion budget. Even non-controversial agencies like the National Institute of Standards and Technology (NIST) would see cuts exceeding 50%. Major programs are slated for elimination, including the NSF’s social science research division and two NIH institutes: the National Institute on Minority Health and Health Disparities and the National Center for Complementary and Integrative Health. Climate change research is a recurring target, with reductions across multiple agencies. The proposal closely mirrors last year’s budget, which Congress largely rejected, suggesting the administration is continuing a political stance rather than pursuing realistic fiscal policy. While Congress may again resist these cuts, the repeated proposals signal sustained pressure on federal science funding. Detailed information limited on specific implementation or long-term planning.

Key Takeaways:
Trump’s 2027 budget proposes deep cuts to science agencies including NSF and EPA
NIH and NIST face significant reductions despite political alignment or low controversy
Climate change research and social sciences targeted for elimination across agencies
Proposal mirrors prior year, indicating political signaling over fiscal reality

Source: Original Article

1. NVIDIA is spotlighting advancements in robotics during National Robotics Week 2026, emphasizing the integration of AI into physical systems across multiple industries. The company highlights progress in robot learning, simulation technologies, and foundation models that are accelerating the development and deployment of intelligent robots. These innovations are enabling robots to transition from virtual training environments to real-world applications in sectors such as agriculture, manufacturing, and energy. NVIDIA’s focus underscores a broader industry shift toward scalable, AI-driven automation, with simulation platforms allowing for safer and more efficient robot training before physical deployment. The emphasis on foundation models suggests a move toward generalized robotic intelligence capable of adapting to diverse tasks. While specific product launches or partnerships are not detailed, the announcement reflects NVIDIA’s strategic positioning in the robotics ecosystem, leveraging its GPU and AI software stack to support next-generation robotic systems. The broader implication is a faster innovation cycle in robotics, driven by improved simulation fidelity and AI model generalization.
   Key Takeaways:
   NVIDIA promotes AI-powered robotics during National Robotics Week 2026
   Robot learning and simulation accelerate real-world deployment across industries
   Foundation models enable more adaptable and intelligent robotic systems
   Source: Original Article

1. Gemma 4 Achieves 2 Million Downloads in First Week, Signals Strong Open Model Adoption
   Google’s Gemma 4 open-weight AI model has surpassed 2 million downloads within its first week of release, marking a significant milestone in the adoption of open-source AI models. This rapid uptake contrasts with Gemma 3’s 6.7 million downloads over a full year and Gemma 2’s 1.4 million since its June 2024 launch, indicating accelerated interest in newer iterations. The model’s performance is being widely tested on consumer hardware, particularly Apple devices, with users reporting efficient on-device inference—such as 40 tokens per second on an iPhone 17 Pro using MLX. Red Hat has also contributed by releasing quantized versions of the 31B parameter model in NVFP4 and FP8-block formats, with instruction-following evaluations available and reasoning and vision benchmarks pending. Gemma 4 has become the top trending model on Hugging Face, reflecting strong community engagement. Its success highlights a broader industry shift toward practical, deployable open models optimized for edge and local inference, especially on Apple Silicon. A keynote event is scheduled in three days from London to further showcase the model’s capabilities.
   Key Takeaways:
   Gemma 4 reached 2 million downloads in its first week
   Strong adoption on Apple hardware signals edge AI momentum
   Red Hat supports deployment with quantized model formats
   Source: Original Article

1. Indian startup Rocket has launched Rocket 1.0, a platform designed to assist entrepreneurs and product teams in defining what to build by generating AI-driven, consulting-style product strategies. Based in Surat, the company aims to address a gap in the current AI landscape, where coding tools have become widely accessible but strategic decision-making remains challenging. The platform integrates research, competitive intelligence, and product development into a unified workflow, producing detailed strategy documents that include pricing models, unit economics, and go-to-market plans. These outputs are delivered in PDF format and resemble traditional consulting reports rather than typical AI coding assistants. Rocket leverages over 1,000 data sources—including Meta’s ad libraries, Similarweb’s API, and proprietary web crawlers—to inform its analysis. While initial testing by TechCrunch confirmed the platform’s ability to generate coherent strategy documents from simple prompts, some insights appeared synthesized from existing public data rather than derived from original research, suggesting users should validate recommendations before implementation. The company offers tiered subscription plans starting at $25 per month for basic application building, scaling up to $250 for advanced features. Human support is available for users encountering issues.

Key Takeaways:
Rocket 1.0 generates AI-powered product strategy reports resembling consulting deliverables
Platform focuses on strategic planning amid rise of accessible AI coding tools
Outputs based on aggregated public data require user validation before use
Subscription pricing ranges from $25 to $250 per month

Source: Original Article

1. OpenAI has proposed a policy framework to address economic disruptions caused by artificial intelligence, including the implementation of taxes on AI-generated profits, the creation of public wealth funds, and the expansion of social safety nets. The proposal aims to mitigate job displacement and rising inequality driven by automation and AI adoption. By taxing AI profits, the plan seeks to generate public revenue that could be reinvested into communities affected by technological change. Public wealth funds would distribute returns from these taxes broadly, potentially providing universal dividends or funding public services. Additionally, enhanced safety nets—such as unemployment support and retraining programs—are recommended to support displaced workers. The framework represents a hybrid approach, combining market-driven innovation with redistributive policies, and comes as governments and institutions worldwide grapple with AI’s socioeconomic implications. While OpenAI’s proposal is not a formal policy, it contributes to ongoing debates about how to manage the transition to an AI-augmented economy. Detailed information on implementation mechanisms or specific tax rates is limited.

Key Takeaways:
OpenAI advocates taxing AI profits to fund public benefits
Proposal includes public wealth funds and expanded worker safety nets
Framework blends capitalist innovation with redistributive economic policies
Policy ideas emerge amid global debate on AI’s economic impact

Source: Original Article

1. Florida’s Fish and Wildlife Conservation Commission (FWC) police have conducted dozens of license plate lookups using Flock’s AI-powered surveillance cameras on behalf of U.S. Immigration and Customs Enforcement (ICE), according to public records. Although ICE lacks a direct contract with Flock, it continues to access the company’s national network of license plate readers through intermediary law enforcement agencies like the FWC. This indirect access raises transparency and accountability concerns, as the FWC’s primary mandate involves wildlife conservation and maritime enforcement, not immigration. Despite Flock’s public statement in May 2025 asserting it does not share data with ICE or other Department of Homeland Security sub-agencies, records show that as of January 2025, thousands of agencies nationwide were sharing camera data with the FWC, which then performed searches for ICE. Flock’s system allows cross-jurisdictional searches, enabling agencies in one state to query data from cameras across the country. This loophole undermines recent reforms aimed at letting local agencies opt out of data sharing with federal immigration authorities. The situation illustrates how surveillance data can be repurposed beyond its original intent, with limited public oversight.

Key Takeaways:
ICE accesses Flock camera data indirectly through state wildlife police
Flock denies direct data sharing with ICE despite evidence of indirect access
Surveillance data repurposing raises transparency and accountability concerns
National lookup tool enables cross-state license plate tracking

Source: Original Article

1. Quittr, an app designed to help users reduce or stop masturbation, faced serious security vulnerabilities related to its use of Google Firebase. At least three independent security researchers identified and reported the issue over several months, beginning as early as September of the previous year. The flaw stemmed from a misconfiguration in Firebase that allowed unauthorized users to authenticate and access backend storage containing user data. One researcher, Kaeden, detailed providing step-by-step guidance to Quittr’s developers, including the need to implement proper security rules, but received no meaningful response despite multiple follow-ups. Another researcher initially alerted the company in September, prompting founder Alex Slater to acknowledge the issue and claim it would be fixed within hours—yet the vulnerability remained unaddressed for months. 404 Media first learned of the issue in January and contacted Quittr multiple times; Slater initially denied the existence of a vulnerability before ultimately resolving it by March 10. The delay in remediation raised concerns about the app’s data protection practices, particularly given the sensitive nature of its user base. Detailed information limited on the full scope of exposed data or number of affected users.
   Key Takeaways:
   Security flaws in Quittr allowed unauthorized backend access for months
   Researchers provided detailed fixes but received delayed or no response
   Founder acknowledged issue early but failed to act promptly
   Vulnerability resolved only after media pressure and repeated outreach
   Source: Original Article

1. A federal appeals court ruled that New Jersey cannot enforce state gambling laws against Kalshi, a prediction market platform registered with the US Commodity Futures Trading Commission (CFTC) as a designated contract market (DCM). The US Court of Appeals for the 3rd Circuit upheld a preliminary injunction in a 2-1 decision, affirming that the CFTC has exclusive jurisdiction over swaps traded on DCMs under the Commodity Exchange Act. The court determined that Kalshi’s sports-related event contracts qualify as swaps, placing them under federal oversight and preempting conflicting state regulations. The ruling rejected New Jersey’s argument that its broader authority to regulate sports gambling applies, emphasizing that the state’s laws directly interfere with federally regulated markets. The case originated when New Jersey issued a cease-and-desist letter to Kalshi, claiming it violated state law by offering unlicensed sports wagers, including on college sports, which are constitutionally prohibited in the state. The decision reinforces federal preemption in financial instruments traded on CFTC-licensed platforms, potentially limiting states’ ability to regulate similar prediction markets. This precedent may influence how other states approach regulation of event contracts and prediction markets.

Key Takeaways:
Federal law preempts state gambling regulations for CFTC-licensed prediction markets
Kalshi’s sports event contracts ruled as swaps under federal jurisdiction
States may face limits in regulating federally designated contract markets
Ruling strengthens CFTC authority over prediction market platforms

Source: Original Article

1. U.S. Health Secretary Robert F. Kennedy Jr. has amended the charter of the Centers for Disease Control and Prevention’s (CDC) Advisory Committee on Immunization Practices (ACIP), granting himself broader authority to select committee members and relaxing membership requirements. The changes were published in the Federal Register following a federal court ruling that temporarily blocked Kennedy’s previous handpicked appointees. In March 2026, U.S. District Judge Brian Murphy ruled that Kennedy’s selections lacked necessary scientific expertise and failed to meet federal standards requiring advisory panels to be “fairly balanced” in representing diverse scientific viewpoints. The judge’s decision suspended ACIP operations and reversed recent policy changes made under Kennedy’s influence, including the removal of recommendations for routine COVID-19 vaccinations and the Hepatitis B birth dose—moves criticized by major medical and public health organizations. The updated charter, part of a biennial renewal process, introduces expanded language that appears to centralize appointment power within the Health Secretary’s office. This revision may enable Kennedy to reconstitute the panel with individuals aligned with his anti-vaccine stance, potentially undermining the committee’s scientific integrity and public health guidance. The move raises concerns about political interference in evidence-based policymaking and could erode trust in national immunization recommendations. Detailed information on specific new membership criteria remains limited, but the structural shift in governance is clear.
   Key Takeaways:
   Kennedy gains greater control over vaccine advisory panel appointments
   Court previously blocked unqualified appointees for lacking scientific expertise
   ACIP operations suspended after controversial policy changes reversed
   Charter changes may enable ideological over scientific appointments
   Source: Original Article

1. NASA’s Artemis II mission achieved a historic milestone as its crewed Orion spacecraft approached the Moon, capturing detailed views of both the near and far sides during a close lunar flyby. On Monday, astronauts aboard the spacecraft observed the Moon for nearly eight hours, with commander Reid Wiseman describing the experience as surreal and beyond current descriptive language. The mission, carrying four crew members—Wiseman, Victor Glover, Christina Koch, and Jeremy Hansen—marks the first human journey to the vicinity of the Moon in over 53 years. Live video feeds from external GoPro cameras showed the Moon expanding in view during final approach, though bandwidth constraints limited resolution. Higher-quality images were expected to be transmitted in the following hours. The crew, trained extensively over three years, prepared for spacecraft operations, emergency scenarios, and scientific documentation, including geology and photography. The Orion spacecraft, named Integrity, has performed nominally since its launch, successfully executing a trajectory that took it behind the Moon. This mission serves as a critical test for future crewed lunar landings under the Artemis program.

Key Takeaways:
Artemis II completes first crewed lunar flyby in 53 years
Orion spacecraft performs well during historic deep space mission
Astronauts trained extensively for operations and scientific observation
Live lunar imagery transmitted despite bandwidth limitations

Source: Original Article

AI processing failed: Connection error.

1. Wisconsin Governor Tony Evers vetoed Assembly Bill 105 on April 3, blocking a proposed law that would have required age verification for accessing websites with more than one-third adult content. The bill mirrored similar legislation passed in numerous U.S. states since 2022. It mandated that users verify their age using methods such as government ID uploads, biometric facial scans, or credit card verification to access sites containing material deemed “harmful to minors,” including depictions of sexual acts or specific body parts. Evers opposed the bill on privacy grounds, stating it imposed an undue burden on adults seeking access to constitutionally protected content. He emphasized concerns about personal data being shared with third parties, including government agencies or data brokers, and highlighted risks of data breaches, identity theft, and blackmail. Although the bill included penalties for improper data retention, Evers argued these could not reverse potential harms once data was exposed. His veto preserves unrestricted access to adult content for Wisconsin residents, distinguishing the state from others with strict age verification laws.

Key Takeaways:
Governor Evers vetoed Wisconsin’s age verification bill over privacy concerns.
The bill required ID or biometric scans for adult content access.
Data security and third-party sharing risks influenced the veto decision.

Source: Original Article