Senate passes $70B immigration bill after overnight vote; Microsoft caught training AI models on unlicensed data; Bitcoin crashes as crypto market tumbles; Meta's AI chatbot hijacks Instagram accounts; Anthropic claims Claude now writes 90% of its own code.
1. Microsoft Trained AI Models on Unlicensed Web Data Despite 'Enterprise Grade' Promises
Microsoft promised enterprise customers 「clean and commercially licensed data」 for its MAI models, but investigation reveals the company trained them on unlicensed web content. This undermines trust as enterprises consider billions in AI infrastructure contracts, raising legal and ethical concerns about data use without permission.
2. Senate Passes $70 Billion Immigration Enforcement Bill After All-Nighter Vote
The Senate voted 51-49 early Friday morning to pass Trump's $69.5 billion budget reconciliation package funding ICE and Border Patrol operations through 2029. The package includes funding for immigration enforcement agencies, representing a major win for the Trump administration's immigration agenda despite Democratic opposition.
3. Bitcoin Crashes Below $62K as Crypto Market Tumbles; Zcash Vulnerability Triggers Selloff
Bitcoin has sunk 50% below its all-time high as the broader crypto market enters a dismal period with no relief-buying in sight. The crash intensified Friday when a serious Orchard Pool vulnerability in Zcash emerged, prompting major holders like Arthur Hayes to liquidate positions amid growing concerns about security across the sector.
4. Meta's AI Chatbot Used to Hijack High-Profile Instagram Accounts by Changing Email Addresses
Attackers discovered a critical security flaw in Meta's AI customer support agent: by simply asking the bot to change email addresses associated with accounts, they gained access to high-profile Instagram accounts. This incident reveals a major blind spot in AI security—the risk extends beyond prompt injection to functional account takeover.
5. Anthropic Claims Claude Now Writes Over 90% of Its Own Code; Pushes for AI 'Pause Button'
Anthropic announced that Claude has achieved a major milestone: the AI now handles more than 90% of the company's internal code generation tasks. Simultaneously, the company is advocating for a global 「AI pause button」 mechanism—a safeguard that could halt self-improving models if they pose risks to humanity, signaling growing concerns about AI autonomy.
6. Anthropic's Mythos Model Powers NSA Offensive Cyber Operations Against China and Iran
Classified intelligence sources reveal that Anthropic's Mythos model is powering NSA offensive cyber operations targeting China and Iran. The disclosure raises questions about AI's role in military and intelligence applications, and whether safety-focused AI companies can ethically support government weaponization of their models.
7. Alphabet Seeking Fresh Capital as AI Data Center Capex Doubles to $190 Billion
Google parent Alphabet is turning to investors to help fund massive AI infrastructure expansion, with capital expenditures expected to reach $190 billion this year—double last year's spending. The company's aggressive buildout reflects intensifying competition to secure GPU capacity and train increasingly powerful AI models, signaling a potential market shift toward larger players with deeper pockets.
8. Cisco Patches Critical SD-WAN Zero-Day Now Actively Exploited in Attacks
Cisco warned Thursday of a high-severity, unpatched zero-day vulnerability in Catalyst SD-WAN Manager (CVE-2026-20245) that is being actively exploited in real-world attacks to achieve root privilege escalation. Organizations running affected versions face immediate risk and must prioritize patches as attackers actively weaponize the flaw.
9. Shell Partners With C3 AI to Automate Predictive Maintenance Across Global Operations
Energy giant Shell is deploying C3 AI agents to shift from basic anomaly detection to fully-automated predictive maintenance across its operations. The partnership marks a major real-world deployment of autonomous AI agents in critical infrastructure, demonstrating enterprise confidence in agentic AI for high-stakes applications despite recent security concerns.
10. Senate Republicans Block FISA Extension; Six GOP Senators Vote Against Surveillance Powers
Six Republican senators unexpectedly voted against advancing a motion to debate extending warrantless surveillance powers under FISA 702, creating a procedural roadblock as the authorities are set to expire next week. Senate Democrats also blocked the extension to protest Trump's appointment of billionaire Bill Pulte as Director of National Intelligence, creating rare bipartisan opposition to the intelligence program.
11. New World Screwworm Livestock Pest Confirmed in Texas—First Case in 60 Years
The U.S. confirmed Thursday that the New World screwworm (NWS) fly—a dangerous livestock pest whose flesh-eating larvae can devastate cattle—has been detected in a calf in south Texas. The discovery marks the return of the pest after 60 years and threatens a $15+ billion livestock industry, triggering emergency quarantine measures across ranching states.
12. Anthropic Expands Mythos Model to 150+ Organizations for Cybersecurity Testing
Anthropic is expanding access to its Mythos model across 150+ organizations in 15 countries specifically for cybersecurity research and testing. The move comes amid growing demand for enterprise-grade AI tools to identify vulnerabilities, though timing raises questions given concurrent revelations about Mythos powering NSA operations.
13. Lebanon Ceasefire Falters as Israel-Hezbollah Fighting Continues; U.S. Mediation Stalls
U.S.-brokered ceasefire negotiations between Israel and Hezbollah in Lebanon are deteriorating as both sides continue military operations. Hezbollah, not party to the formal talks, rejected ceasefire conditions as a 「virtual surrender,」 while Iran-backed forces maintain attacks, signaling deepening Middle East instability as the broader U.S.-Iran conflict shows no signs of resolution.
14. Apple Agrees to Reveal India Revenue to Avoid Massive $38 Billion Fine
Apple has agreed to disclose its revenue figures in India to Indian authorities, narrowly avoiding a potential $38 billion penalty—one of the largest fines ever threatened against the tech giant. The settlement comes as India aggressively pursues higher taxes from foreign tech companies, part of broader efforts to increase revenue extraction from multinational corporations operating in the country.
15. Amazon Shifts Prime Day to June 23, Breaking from Traditional July Slot
Amazon has announced that Prime Day 2026 will occur on June 23, marking a historic shift from its traditional July timing. The move aims to kick off summer shopping earlier and capture holiday season preparations sooner, reflecting changing consumer shopping patterns and Amazon's strategy to maximize engagement during shifted retail cycles.
16. MacBook Neo Offers Colorful Customization With Official Replacement Parts
Apple's MacBook Neo—its cheapest and most repairable laptop in years—enables users to customize its design with colorful official replacement parts. This marks a significant shift in Apple's approach to consumer hardware, allowing owners to swap panels and components in all four available colors, combining affordability with unprecedented repairability.
17. Huawei Chips Successfully Refine DeepSeek AI Model, Marking Leap in China's AI Independence
A research team including Huawei has successfully used Huawei's Ascend 910C chips to complete post-training for the DeepSeek-V4-Pro model. The achievement represents a major milestone in China's efforts to reduce dependence on foreign semiconductors for AI development, demonstrating that domestic chip technology can handle critical AI workloads previously thought to require Western hardware.
18. China, Hong Kong Users Cannot Access SpaceX IPO Website Following Blockade
Users in China and Hong Kong are unable to access SpaceX's IPO documents and official website, indicating a likely government-imposed digital blockade. The move reflects geopolitical tensions around SpaceX's record $75 billion IPO filing and broader restrictions on U.S. technology companies' information access in mainland China and Hong Kong markets.
19. US-Iran War Pushing Millions Into Food Crisis, UN Warns
The UN World Food Programme warns that the escalating U.S.-Iran conflict is driving substantial food price increases across multiple countries, pushing millions into food insecurity. The organization cautions that the regional instability threatens global food supplies and humanitarian access, with ripple effects extending far beyond the immediate conflict zones.
20. Actor James Handy From 'Top Gun: Maverick' Stabbed to Death; Girlfriend's Son Arrested
James Handy, a veteran character actor known for appearances in 「Top Gun: Maverick」 and numerous television shows, was fatally stabbed. The Los Angeles Police Department arrested his girlfriend's son in connection with the killing. The death marks a tragic loss for the entertainment industry and raises concerns about domestic violence and household disputes.