Sift20

1. Trump Says Iran Deal 'Largely Negotiated,' Plans to Reopen Strait of Hormuz

President Trump announced Saturday that a peace deal with Iran is 「largely negotiated」 and would include reopening the Strait of Hormuz, following calls with Arab leaders. The agreement marks a potential turning point in the three-month conflict that has devastated global energy markets and sent oil prices to four-year highs.

2. Ebola Death Toll Surges Past 200 in DR Congo; 10+ African Nations at Risk

The death toll from the Ebola outbreak in the Democratic Republic of Congo has climbed to 204, with Uganda and at least 10 other African countries now flagged as at-risk. The rapid spread of the rare Bundibugyo strain has prompted the African CDC to escalate emergency response and prompted some international travel restrictions.

3. Pakistan Train Bombing Kills at Least 24 Military Personnel in Balochistan

A suicide bomber detonated an explosive-laden vehicle near a passenger train in Quetta, Pakistan on Sunday, killing at least 24 people, mostly military servicemen, in one of the deadliest attacks in the volatile southwestern province. The bombing underscores ongoing security challenges in the strategically important region.

4. Russia Unleashes Hypersonic Oreshnik Missile in Massive Attack on Kyiv

Russia confirmed using its Oreshnik hypersonic ballistic missile—which travels over 10 times the speed of sound—in a coordinated drone and missile barrage on Kyiv Sunday morning, killing at least four and injuring dozens. The attack marks an escalation in Russia's use of advanced weaponry against Ukraine's capital.

5. Claude AI Discovers Novel Scaling Algorithms Humans Might Not Have Designed

Anthropic researchers used Claude Code to autonomously discover new AI scaling algorithms that differ from conventional human-designed approaches. The breakthrough demonstrates AI's potential to uncover counterintuitive solutions in machine learning optimization—a capability that could accelerate AI development.

6. Anthropic Warns Claude Finds Bugs Faster Than Developers Can Patch Them

Anthropic has warned that Claude's bug-finding capabilities in its latest Mythos Preview may be outpacing human developers' ability to patch vulnerabilities. The concern raises important questions about AI-assisted security testing and the potential for automated systems to discover exploits faster than remediation can occur.

7. Trump Administration Scraps AI Testing Order After Pressure From Musk, Zuckerberg

President Trump abruptly withdrew a planned executive order on AI safety testing after pressure from tech billionaires including Elon Musk and Mark Zuckerberg. The reversal exposes internal White House divisions over AI regulation and highlights the influence of major tech figures on government policy.

8. Nuro Positioning Itself as Robotaxi 'Second Mover' Against Waymo's Dominance

Robotaxi startup Nuro argues that being a 「second mover」 in the autonomous vehicle market gives it strategic advantages over market leader Waymo, which operates over 3,000 driverless cars across 10+ US cities. Nuro's approach focuses on learning from Waymo's challenges while building differentiated technology.

9. AI Model Selection Default Settings Hide Risks; Users Should Choose Consciously

New research warns that leaving AI model selection on default in tools like Copilot and Gemini can expose users to data bias and privacy risks. The analysis reveals that default model choices may not align with user needs or values, requiring active decision-making about which AI systems to use.

10. Middle East War Tests Gulf Region's Ambitions to Become AI Hub

Attacks on data centers and persistently high energy prices from the Iran conflict are forcing major recalculations for data center operators eyeing the Gulf as an AI hub. The geopolitical risks and infrastructure vulnerabilities are dampening investment plans that had positioned the region as a future AI powerhouse.

11. Anthropic May Continue NSA Supply Deal Despite Pentagon Supply Chain Risk Flag

Despite the Pentagon flagging Anthropic as a potential supply chain risk, the AI safety company may maintain its agreement to provide Claude to U.S. intelligence agencies. The tension reflects broader challenges in governing AI security and vendor relationships within government institutions.

12. Scammers Abusing Microsoft Internal Account to Send Legitimate-Looking Spam

Threat actors have compromised an internal Microsoft account and are using it to distribute phishing links and spam that appear authentic, bypassing standard email security filters. The breach highlights vulnerabilities in how legitimate organizational accounts can be weaponized for mass social engineering attacks.

13. Discord Enables End-to-End Encryption for All Voice and Video by Default

Discord has rolled out default end-to-end encryption for all voice and video communications on its platform, significantly enhancing user privacy. The move makes encrypted communications the standard rather than an opt-in feature, strengthening user protection against eavesdropping.

14. French Open 2026 Begins Amid Paris Heat Wave; Sinner and Djokovic Compete

The 125th French Open (Roland Garros) has commenced in Paris under scorching heat conditions, with top-seeded players including Jannik Sinner, Novak Djokovic, Aryna Sabalenka, and Coco Gauff competing for Grand Slam glory. Extreme temperatures are expected throughout the tournament weekend.

Trending: french open

15. Deepseek Makes 75% Discount Permanent; Output Tokens 34x Cheaper Than GPT-4.5

Chinese AI startup Deepseek has permanently locked in its aggressive 75% pricing discount, making its API tokens significantly cheaper than OpenAI competitors. The pricing strategy signals Deepseek's intent to disrupt the global AI market and challenges assumptions about AI model profitability.

16. Scientists Discover Hidden Aging Driver; Menin Supplement Reverses Brain Decline

Researchers have identified a brain protein called Menin whose declining levels trigger inflammation, memory problems, and bone loss associated with aging. Preliminary studies show that a simple supplement may reverse brain decline and slow aging processes, opening potential therapeutic pathways.

17. Apple Registers 'Gen AI' Subdomain Ahead of WWDC Keynote

Apple has registered a new subdomain focused on generative AI, signaling major AI announcements at next month's Worldwide Developers Conference keynote. The move suggests Apple plans significant AI feature expansion following previous 「Apple Intelligence」 rollouts.

18. FBI Seeks 'Near Real-Time' Access to License Plate Readers Nationwide

The FBI has requested expanded surveillance capabilities through license plate reader data, seeking near real-time access to tracking information across the United States. The proposal raises privacy concerns and highlights ongoing debates over government surveillance infrastructure.

19. Cannes Film Festival 2026: Neon Wins Seventh Consecutive Palme d'Or with 'Fjord'

Neon Distribution secured its seventh consecutive Palme d'Or at Cannes with director Cristian Mungiu's 「Fjord,」 starring Renate Reinsve and Sebastian Stan. The unprecedented streak underscores the indie distributor's dominance in securing top festival honors and securing prestige films.

20. Aeronautical Engineering Principle Overturned: Roughness Sometimes Reduces Drag

Scientists have discovered that the long-held aeronautical principle that smoother surfaces always reduce drag is not universally true. In certain conditions, roughness can actually decrease aerodynamic drag, overturning conventional wisdom and potentially transforming aircraft and vehicle design.