Top HN Weekly Digest · W21, May 18-24, 2026

The Trump administration has issued a new policy requiring most green card seekers to leave the United States and apply through consulates abroad, limiting "adjustment of status" within the country to only extraordinary circumstances. [src]

Commenters argue that requiring green card seekers to apply from abroad is a "malevolent" disruption of the only practical path for skilled workers, who cannot realistically maintain US employment while waiting years for uncertain consular processing [0][3]. While some note that leaving a country to renew a visa is common internationally [9], others contend this policy is a cynical attempt to restrict legal immigration by reinterpreting "adjustment of status" as a discretionary grace rather than a standard procedure [2][7]. The discussion highlights a deep frustration with the complexity and perceived hostility of the legal system, with users warning that these hurdles threaten the US technological lead and may inadvertently turn legal residents into "illegal" immigrants [5][6][8].

An OpenAI model has successfully disproved a central conjecture in discrete geometry, marking a significant milestone in the application of artificial intelligence to complex mathematical problem-solving. [src]

The successful disproof of a discrete geometry conjecture has sparked debate over whether LLMs are merely "recombining" training data or performing genuine discovery, with some arguing that even human mathematical breakthroughs often involve unfolding truths already implicit in existing axioms [0][1]. While some mathematicians are optimistic that these tools can help manage the "exploding complexity barrier" of modern research [2], others contend that LLMs remain "permutation machines" incapable of the artistic "creation" required for paradigm-shifting leaps like calculus [3][7][8]. Critics also point out that as AI achieves "PhD-level" milestones, skeptics frequently move the goalposts to demand genius-level innovation [5], while some professionals express concern that such progress may eventually render lifelong human expertise obsolete [6].

Andrej Karpathy, a prominent AI researcher and former founding member of OpenAI, has announced that he is joining the AI safety and research company Anthropic. [src]

The discussion regarding Andrej Karpathy joining Anthropic is divided between those who see him as a top-tier talent who strengthens the pre-training team [1][9] and skeptics who view the move as a "celebrity hire" or marketing stunt intended to boost IPO value [2][3][8]. While some users criticize his frequent job-hopping, others point out that his five-year tenure at Tesla is significant for the tech industry [0][6]. There is also a broader debate about Anthropic’s trajectory, with some users praising their safety-conscious culture [1] while others fear the company is becoming an "industry tornado" that prioritizes hype over product merit [7][8].

Flipper Devices has announced Flipper One, an ambitious open-source Linux "cyberdeck" and network multi-tool featuring a modular hardware design and a dual-processor architecture. The team is seeking community assistance via a new Developer Portal to help refine its custom OS, UI framework, and mainline Linux kernel support. [src]

The Flipper One announcement sparked a debate over its writing style, with some users dismissing the text as "AI slop" [0][8], while others argued that such cynicism is becoming a tiresome distraction from actual content [1]. The author clarified that the text was a human-written draft in Russian and English polished by editors, not generated by AI [5][9]. Beyond the prose, commenters expressed confusion over the specific "help" requested [2], eventually identifying the project as a call for FOSS contributors to assist with hardware reverse-engineering and open-source driver development [6][7]. While some praised the ambitious scope and "all in-tree" source goal [3], others questioned the form factor, suggesting a full QWERTY keyboard and x86 architecture would be more practical for mobile development [4].

Five major European mobile payment providers, including France's Wero and Spain's Bizum, are uniting to launch an independent, interoperable payment network for 130 million users starting in 2026 to challenge the dominance of Visa and Mastercard. [src]

Users generally praise the shift toward sovereign payment systems like Wero and iDEAL for improving security by eliminating the need to share sensitive card data with merchants [0]. While some argue that government-backed systems like Brazil’s PIX offer superior functionality and national autonomy [1][2][8], others warn that these platforms lack the robust consumer fraud protections and chargeback mechanisms provided by Visa and Mastercard [7]. There is also significant skepticism regarding the fragmented nature of European apps and the slow timeline for commercial adoption [4][5], alongside a debate over whether central bank control represents a democratic safeguard or a move toward authoritarianism [3][6].

A California jury unanimously ruled against Elon Musk in his lawsuit against OpenAI and Sam Altman, finding that his claims regarding the company's shift to a for-profit model were filed after the statute of limitations had expired. [src]

Elon Musk lost his lawsuit against OpenAI primarily because the jury determined he waited too long to file, exceeding the three-year statute of limitations [0]. Commentators noted that Musk’s own past emails supporting a for-profit transition and his attempts to merge OpenAI into Tesla undermined his "betrayal" narrative and suggested "unclean hands" [1][2]. While some reflect on the alternate history where Musk might have controlled the AI frontier, others view the lawsuit as a reactionary move following the success of ChatGPT and his own failed attempts to acquire the company [1][4][7].

Google is redesigning its iconic search box to integrate Gemini AI, shifting the platform from a list of links toward a conversational interface that provides direct answers and synthesized information. [src]

The integration of AI into Google Search has sparked significant concern regarding "Google Zero," a scenario where the search engine ceases to drive traffic to external websites, leading site owners to question the value of allowing crawlers at all [0][2]. Users report frequent inaccuracies and "bullshit answers" that complicate professional work and potentially endanger users seeking medical or financial advice [3][6][8]. While some have already shifted their habits toward LLMs or alternative search engines, there is a strong consensus that these AI summaries often present "random stuff" as ground truth while lacking the essential primary sources required for factual reliability [1][4][5][9].

Google has introduced Gemini 3.5 Flash, a high-speed, cost-efficient AI model designed for low-latency performance and high-volume tasks. [src]

The release of Gemini 3.5 Flash has sparked significant concern over its pricing, which represents a 3x to 6x increase over previous Flash models and positions it closer to the cost of older "Pro" versions [0][7]. While some users praise Google’s focus on optimizing smaller models [6], others argue that these rising costs make AI increasingly inaccessible to individuals and suggest that serving LLMs profitably remains a major challenge [2][4][9]. Early testing shows mixed results: the model demonstrates impressive reasoning capabilities for complex SVG generation, yet it can still struggle with anatomical logic in images and carries a high per-request cost for long outputs [1][3].

The author argues that AI companies and users engage in large-scale unauthorized plagiarism by training models on uncompensated content and profiting from generated results, citing a personal experience where a competitor used ChatGPT to copy their tutorials and outrank them in search results. [src]

The debate centers on whether the massive scale of AI training constitutes a qualitative shift that distinguishes it from individual learning or fair use [1][7]. While some argue that intellectual property is an outdated concept and that AI could finally dismantle restrictive copyright laws [0][4], others contend that removing ownership disincentivizes creation and unfairly exploits content providers who fund the very data AI consumes [2][3]. This tension is further complicated by accusations of hypocrisy regarding past support for ad-blocking and piracy, alongside concerns that websites may soon be forced behind logins to survive [2][9].

Human rights organizations are condemning Meta for geo-blocking the Facebook and Instagram accounts of NGOs and activists in Saudi Arabia and the UAE following government requests to restrict content under local cybercrime laws. [src]

The discussion highlights a tension between social media's original promise to spread democracy and its current role as a tool for state-level censorship and propaganda [1][4]. While some argue that platforms are forced to comply with local laws to avoid being banned entirely [5][7], others contend that the "privatized profits, socialized harm" model incentivizes companies to prioritize revenue over human rights [0][3]. Users are increasingly skeptical of large-scale networks, though they struggle to find viable alternatives that balance community connection with protection against state influence [6][8].