Interesting Engineering

Samsung Group is expected to announce a 1,000 trillion won ($648 billion) investment plan in South Korea over the next 10 years, according to Maeil Business Newspaper. The announcement is set for Monday at a meeting with President Lee Jae Myung, where top executives from Samsung Electronics and SK Hynix will also present investment plans. The spending will cover AI data centers, chip factories, batteries, and displays, including a potential 300 trillion won push to build semiconductor facilities in South Korea's southwestern region — outside the current concentration of chip manufacturing around Seoul. The plan reflects South Korea's effort to convert surging AI-driven chip demand into broader national economic growth. SK Hynix and Samsung may need to accelerate projects originally planned for the 2040s to the mid-2030s, as AI memory demand is growing faster than available space, power, and water in the Seoul region allow. The initiative carries political dimensions: President Lee won approximately 85% of the vote in Gwangju and neighboring South Jeolla in the 2025 election, and opposition lawmakers have accused his government of directing investment to ruling party strongholds for political reasons. Samsung and SK Hynix declined to comment.

#Samsung #SouthKorea #AI

17 minutes ago | [YT] | 3

Interesting Engineering

The Trump administration has asked OpenAI to limit the release of its newest model, GPT-5.6, to a select group of close partners rather than making it publicly available, according to The Information. CEO Sam Altman reportedly told staff that the government would be "approving access customer by customer" during a preview period, with a broader release possible "a couple of weeks later" if the limited rollout goes well. The request came from the Office of the National Cyber Director and the Office of Science and Technology Policy, and OpenAI's staff reportedly worked closely with the government on the upcoming release. The move mirrors Anthropic's approach with Claude Mythos, which was restricted to a small group of partners through Project Glasswing due to its advanced cybersecurity capabilities. Earlier this month, Trump signed an executive order directing AI companies to voluntarily submit new models for government testing before public release — a shift from the administration's originally stated "hands-off" approach to AI. The core concern with frontier cyber models is their ability to identify and exploit software vulnerabilities at speeds no human analyst could match.

#OpenAI #AI #NationalSecurity #Trump #TechPolicy

1 hour ago | [YT] | 17

Interesting Engineering

This black fuel keeps global trade moving—but its hidden costs are staggering. Watch the full video to find out why. ‪@ie-explains‬

17 hours ago | [YT] | 17

Interesting Engineering

The real story goes far beyond the crashes. Watch the full video to see why the world's most powerful bombers are reaching a turning point. ‪@militarymechanicsie‬

19 hours ago | [YT] | 3

Interesting Engineering

A group of Hollywood stars including Cate Blanchett, Meryl Streep, Emma Thompson, Helen Mirren, Kristen Stewart, George Clooney, Viola Davis, Tom Hanks, and Javier Bardem have joined a new nonprofit platform called rslmedia.org, designed to give artists control over how their identity is used by AI. The tool works as a consent registry: artists verify their identity and specify their consent level using a three-tier color system — green for free use, yellow for conditional use such as payment, and red for prohibited use. The database is designed to be machine-readable, allowing AI companies to check consent at scale. The platform was co-founded by Blanchett alongside Nikki Hexum, Doug Leeds, and Eckart Walther, and is built on the principle that human identity — including face, voice, movement, and ideas — is a form of intellectual property. Blanchett presented the initiative at the European Parliament on Tuesday, alongside filmmaker Steven Soderbergh, who described it as "a persuasive mechanism to do the right thing in a simple and elegant way — not a law, not a restriction." The launch comes as AI-generated likenesses of celebrities and public figures have become increasingly common and difficult to regulate.

#AI #CateBlanchett #Hollywood

20 hours ago | [YT] | 39

Interesting Engineering

The US government declared an end to its response to the hantavirus outbreak aboard the cruise ship MV Hondius on June 24, after the final 42-day monitoring period wrapped up with no American passengers contracting the virus. Health Secretary RFK Jr. credited federal efforts for preventing "sustained transmission" — though no Americans brought the virus home in the first place. Significant questions remain about the administration's quarantine approach, which was stricter than protocols used in Chile and Argentina where the Andes virus is endemic, stricter than WHO recommendations, and stricter than what the CDC's own expert advised. In one notable case, passenger Angela Perryman sought to quarantine at home in Florida. CDC quarantine medical reviewer Michael Bell assessed her case and firmly recommended she be allowed to return home, writing it was "consistent with the level of transmission risk associated with Andes virus infection." Four days later, Kennedy signed an order keeping Perryman in Nebraska's federal quarantine unit — with no explanation for overruling Bell. At a press briefing following the announcement, CDC officials dodged all questions about the case and the justification for the strict protocols. The global outbreak tally stands at 13 cases and three deaths, with WHO expected to declare the outbreak over by July 2 if no new cases emerge.

#Hantavirus #PublicHealth #RFKJr

23 hours ago | [YT] | 85

Interesting Engineering

Anthropic sent a letter to the US Senate Banking Committee accusing Alibaba of carrying out "the largest known distillation attack on Anthropic to date," CNBC confirmed. The letter, dated June 10 and addressed to Senators Tim Scott and Elizabeth Warren, states that operators affiliated with Alibaba and its AI lab conducted 28.8 million exchanges with Anthropic's models using approximately 25,000 fraudulent accounts between April 22 and June 5. Distillation is an AI training technique in which a smaller, weaker model is built using outputs from a larger, more capable one — effectively transferring capabilities without authorization. Anthropic said Alibaba "ignored the Trump Administration's warnings" despite a White House memo in April pledging to help AI companies detect and coordinate against industrial-scale distillation. The accusation comes two months after Anthropic identified three other industrial-scale distillation campaigns from DeepSeek, Moonshot, and MiniMax. Alibaba did not immediately respond to a request for comment. The letter also arrives as Anthropic is navigating a separate dispute with the Trump administration over export controls on its Fable 5 and Mythos 5 models, which remain offline pending resolution.

#Anthropic #Alibaba #AI #Cybersecurity #TechNews

1 day ago | [YT] | 43

Interesting Engineering

On June 24-25, two significant seismic events occurred within roughly 25 minutes of each other on opposite sides of the planet. In Venezuela, a magnitude 7.2 earthquake was followed just 39 seconds later by a magnitude 7.5 in nearly the exact same location — a rare phenomenon known as a "doublet" earthquake. The back-to-back strikes collapsed buildings in Caracas, killed at least 32 people, injured hundreds, and prompted acting President Delcy Rodriguez to declare a state of emergency. Simón Bolívar International Airport was closed due to damage. Approximately 25 minutes after the Venezuelan doublet, a separate magnitude 6.9 earthquake struck off the northeastern coast of Japan's Iwate Prefecture. No tsunami warning was issued and no injuries were reported. Seismologists note there is no evidence the two events are connected — significant earthquakes occur regularly around the world. What drew attention was the tight timing: large doublet earthquakes are uncommon, and the coincidence of another major quake on the other side of the world within the same half-hour window is statistically unusual, even if not literally once-in-a-millennium.

#Earthquake #Venezuela #Japan #NaturalDisaster #Science

1 day ago | [YT] | 33

Interesting Engineering

Slash, a San Francisco-based fintech company, revealed on X that one of its employees ran up an $81,267 AI coding bill building a video game called "Brainrot Shooter" — a bare-bones first-person shooter set in a Minecraft-like landscape where players shoot characters with internet meme names like "skibidi toilet." The employee, Nicolas Brilliante, Slash's head of strategic verticals, said the spending was "a genuine accident — I underestimated my own ability." Slash leaned into the situation with humor, posting: "Pls play it so we can write this off as a marketing expense." The incident comes as companies across the industry are rethinking runaway AI spending. Uber, Coinbase, and Walmart have all introduced caps on employee AI usage, with Walmart specifically citing unnecessary and repetitive "vibe coding" as a concern. Prediction market Polymarket noted that Slash was "forced to roll back its AI coding push" after the incident. Brilliante himself mused: "Am I going to become a case study for how AI spend can get out of control?"

#AI #Fintech #VibeCode #TechNews

1 day ago | [YT] | 59

Interesting Engineering

Engram, an AI memory startup founded in October 2025, has raised $98 million from General Catalyst, Kleiner Perkins, Sequoia, and Anthropic-bound OpenAI co-founder Andrej Karpathy. The 13-person company claims its models can match or outperform frontier AI labs using up to 100 times fewer tokens — the unit of compute that determines the cost of running AI queries. Engram describes itself as the "learned memory" of AI: rather than processing everything from scratch, its models recall organization-specific workflows and context to anticipate questions and deliver cheaper, more targeted responses. The startup's customers already include Microsoft, Notion, and legal AI startup Harvey. The raise comes as corporate AI spending faces growing scrutiny — newer, more capable models are proving more expensive than their predecessors, challenging the assumption that scaling AI would drive costs down. CEO Dan Biderman, who holds a PhD in computational neuroscience from Columbia, said the goal is to build "a layer of intuition that humans have, and current models don't." Engram plans to use the funding to expand compute capacity and hire talent.

#AI #Startups #Funding #TechNews #MachineLearning

1 day ago | [YT] | 66