Blinken Says Iran Nuclear Issue and Lebanon Ceasefire Are Key Focus of Talks

**June 21 — Palantir CEO’s Controversial AI-Politics Remarks Spark Industry & Political Stir** The head of tech firm Palantir has grabbed widespread attention for explosive recent comments directly linking artificial intelligence’s economic disruption to U.S. party politics, focusing on highly educated white-collar workers — the core voting bloc of the U.S. Democratic Party. In remarks, he claimed AI will "significantly weaken" the economic influence of highly educated voters. He warned anyone who believes this economic shift won’t trigger major political fallout is "no different from being in a madhouse," adding, "We cannot allow this technology to upend humanity’s democratic voter base." These comments, which explicitly tie AI’s economic impacts to party dynamics, have sparked debate across both the tech and political sectors. As of press time, Palantir has not released further clarification on the remarks. --------------------------------- Click the original article link

[June 21] – GPU vendor Tinygrad announced today, citing reliable industry sources, that its GLM 5.2 model achieves an inference speed of 120 tokens per second when run on two interconnected Blackwell architecture TinyBox units. The configuration carries a price tag of $150,000, with buyers able to select either two standard TinyBox units or a single TinyBox Pro unit — both setups deliver identical performance. Tinygrad’s core selling point for this offering is its “one-time purchase, no recurring cloud fees” model for private deployments, directly competing with on-demand cloud-based inference services. Notably, the report has not been officially verified by the GLM team, and Tinygrad has not disclosed additional technical details about the system. To stay updated on global AI hot topics around the clock, click the original article link to join BlockBeats’ Lark AI News Channel.

June 21 — According to a Financial Times report, index provider MSCI has slapped its lowest Environmental, Social, and Governance (ESG) rating — a CCC — on SpaceX, right ahead of the space firm’s record-breaking $75 billion initial public offering (IPO) this month. MSCI noted this rating puts SpaceX “behind its industry peers, thanks to high-risk exposure and its failure to manage major ESG risks.” Frédéric Ducoulombier, director of the Climate Research Institute at EDHEC Business School, weighed in: “The company’s poor controversy assessments, extremely weak governance scores, and lower overall ESG rating shouldn’t come as a shock. For public market investors, this is basically a governance horror story.”

According to The Associated Press, U.S. Vice President JD Vance arrived in Switzerland on June 21 to lead U.S. negotiations with Tehran over Iran’s nuclear program. Ahead of the talks, Vance outlined his top priorities for the Iran negotiations: establishing a high-level, politically driven framework for discussions, making progress on the nuclear issue, and pushing for a ceasefire in Lebanon. “We will put in place a politically led framework at the highest level, while technical teams continue working on-site,” he stated. He added that he will only remain in Switzerland for “one to two days” during this trip, but hopes to secure progress on Iran’s nuclear material handling. Vance also emphasized that pushing a ceasefire in Lebanon is another key priority, coming amid recent renewed missile attacks from Israel in the region. “This is a situation we must keep managing. These two issues are our focus. I believe the Iranian side will also have issues they wish to discuss,” he said.

June 21 — As AI-driven security tools (like Mythos) grow in popularity, the cost of smart contract audits in the crypto space is dropping rapidly. Researchers believe this could fundamentally shift the industry’s longstanding standard of “reasonable due diligence” before a project launches. A new report notes that AI security systems can detect vulnerabilities at a far lower cost than traditional manual audits and support continuous code review, moving away from the old “one-time audit” model to real-time, ongoing monitoring. Industry experts say this shift could push basic security audit costs to nearly zero. Security specialists explain AI not only catches standard code vulnerabilities but also simulates attacker behavior to test how code performs in real adversarial environments—significantly boosting the efficiency of finding flaws. However, they emphasize AI still struggles to identify complex risks such as economic model flaws, permission misuse, and social engineering attacks.

June 21. Serenity’s article points out that the Trump administration may be realizing the core leverage of its trade war strategy lies not in traditional goods exports, but in frontier supply chains centered on quantum computing, artificial intelligence, and robotics. However, key nodes of these critical supply chains are currently mostly controlled by U.S. allies: Dutch firm ASML monopolizes extreme ultraviolet lithography machines; Japanese companies TOWA, Ibiden, and Ajinomoto dominate advanced packaging key materials and equipment; while Taiwan, South Korea, and the European Union each hold irreplaceable positions in different stages of semiconductor manufacturing. Serenity states that imposing tariffs on allies has actually weakened the U.S.’s ability to integrate these vital resources. In the current context, if the Trump administration hopes to secure more favorable terms in trade negotiations, repairing relationships with allies and transforming the global supply chain into s