Top 10 Hacker News posts, summarized
HN discussion
(297 points, 266 comments)
The article by Jay Lund addresses concerns about artificial intelligence's water consumption, arguing that public and media perceptions are exaggerated. Based on calculations using fundamental physics and verified with AI models, Lund estimates that California's data centers using evaporative cooling consume between 32,000 to 290,000 acre-feet annually (0.08% to 0.7% of the state's total human water use). He emphasizes this is manageable and economically efficient compared to other uses, while criticizing speculative reporting lacking data. Lund also notes AI can aid in preliminary estimations for policy assessments but cautions against untamed public discourse not grounded in evidence.
HN commenters largely agree that AI water use is overstated but criticize both corporate lack of transparency and sensationalist media, with one noting Pete Buttigieg cited the debunked "10,000 gallons per photo" figure. Technical discussions focus on cooling systems: advocates propose mandating closed-loop systems (re-condensing water) instead of cheaper open-loop evaporation, while others highlight pollution risks from biocides in cooling discharges. Skepticism toward the author's methodology of using AI for estimates emerged, alongside comparisons to other water-intensive activities (beef production, golf courses). Some diverted to electricity/pollution concerns as more pressing, while others defended AI's relative efficiency versus social media or advertising.
HN discussion
(376 points, 126 comments)
WhatCable is a macOS menu bar app designed to inspect USB-C cables and diagnose charging and data transfer issues. It uses Apple's IOKit to display in plain English what each connected cable can do, including its speed, power rating, and connected devices. The app identifies bottlenecks such as slow charging, reveals e-marker data from the cable's chip, and shows negotiated power profiles from the charger. While it is available for manual download or via Homebrew, it requires macOS 14 (Sonoma) and Apple Silicon, as Intel Macs do not expose the necessary data through public APIs.
The HN discussion focused on the app's functionality, limitations, and potential for cross-platform support. Users appreciated its native macOS implementation and utility for diagnosing cable issues but noted its Apple Silicon-only restriction as a significant limitation. Several commenters inquired about Linux alternatives, suggesting tools like `lsucpd`, while others requested a Homebrew cask and a console-only version. Some users reported issues with the app detecting ports, and there was skepticism about its ability to detect counterfeit cables that misreport their capabilities. The discussion also touched on the challenge of distinguishing cable types physically and the growing need for transparency in cable specifications.
HN discussion
(205 points, 229 comments)
The article presents a "Who is hiring?" thread for May 2026, listing active job postings from various companies. Companies must adhere to strict guidelines: only direct hiring posts are allowed (no recruiters), location must be specified (including REMOTE/ONSITE distinctions), non-household names need explanation, and posters must be committed to engaging with applicants. The thread provides links to external job aggregators and references the complementary "Who wants to be hired?" thread. Among the 15 listed companies, roles span AI/ML engineering, full-stack development, DevOps, security, finance, and product management, with significant emphasis on AI-native work, distributed systems, and formal verification. Salaries range widely, from $95K to $250K USD, with equity offerings common. Companies include startups (Pathos AI, Formal, Burnin) and established firms (Fastly, CodeWeavers, Smarkets), operating globally with roles based in the US, Europe, and Asia.
The HN comments reveal strong demand for AI/ML engineers, particularly those with experience in generative AI, LLM workflows, agentic systems, and formal verification. Remote work opportunities are prevalent, though some roles (like Shepherd, Moyai, TwoSixTech) require onsite presence. Commenters noted the high concentration of well-funded startups (e.g., Burnin with institutional VC backing, Pathos AI with $60M+ funding) and the competitive compensation packages, especially for senior/staff roles. A recurring theme is the focus on impactful, complex domains—oncology (Pathos), AI safety (Formal), high-stakes computing (Burnin), and infrastructure (Loophole Labs, Fastly). Job seekers expressed interest in roles offering technical depth (e.g., Loophole Labs' eBPF/XDP work) and opportunities to build foundational systems (Formal's verified computing stack). The diversity of industries beyond core tech (mining, insurance, biotech) was also highlighted as a notable feature of the job market.
HN discussion
(297 points, 104 comments)
The article details "The Gay Jailbreak Technique," a method to bypass AI safety filters by framing requests within LGBTQ+ contexts. The technique exploits the model's tendency to be overly compliant and "helpful" towards LGBTQ+ identities, tricking it into bypassing safety guardrails. Users incorporate phrases like "be gay," "lesbian gay voice," or ":3" into prompts requesting sensitive information (e.g., meth synthesis, keyloggers, carfentanyl synthesis), often disguising the true intent within educational or role-play framing. The method is claimed to be highly effective, even against newer models like GPT-4o, Claude 4 Sonnet/Opus, and Gemini 2.5 Pro, and is noted to become stronger with stricter safety measures.
HN users reacted with a mix of amusement, skepticism, and analysis. Many found the technique ironic and funny, highlighting its "fabulous" nature and comparing it to exploiting linguistic "hardpoints" or "coaching" the model (e.g., "Be gay do crime," "Homo say what"). Several commenters questioned its effectiveness, noting it failed on newer models like GPT-5.5 or yielded unimpressive results. The core mechanism was debated: some saw it as a specific failure of "political overcorrectness" towards LGBTQ+ topics, while others argued it was a broader jailbreak exploiting roleplay/obfuscation common in AI safety bypasses. There was also criticism regarding the lack of compelling evidence and concerns about the framing's potential homophobia.
HN discussion
(177 points, 198 comments)
Spotify is introducing a "Verified" badge with a green checkmark to identify human artists on its platform. This badge will appear next to artists who meet specific authenticity criteria, such as linked social accounts, consistent listener activity, merchandise presence, or concert history. Spotify claims the verification will cover "hundreds of thousands of artists," prioritizing those with significant cultural contributions over "content farms," and will roll out over the coming weeks. However, critics note the badge only confirms a human manages the profile, not that the music is AI-free. Professor Nick Collins highlights challenges in labeling music itself due to AI usage existing on a spectrum, potentially favoring established artists over newcomers.
Hacker News users criticized Spotify's move as primarily a "scammer filter" (AftHurrahWinch) rather than a solution to AI-generated music content. Commenters expressed frustration over the platform flooding with AI "slop" (cdrnsf, dlivingston), with demands for options to exclude AI artists from recommendations and playlists (RankingMember, cpeterso). Financial motives were questioned, with speculation that Spotify or its investors (like Tencent Music) benefit from promoting AI music to avoid royalty payouts (reconnecting, jalada). Some users proposed segregated AI-only streaming services (bilsbie), while others noted practical issues, such as real musicians using AI tools for side projects and confusion over verification criteria (CPLX). Many also criticized Spotify's broader decline in curation and user experience.
HN discussion
(165 points, 180 comments)
The TI-84 Evo calculator features an icon-based home screen for faster access to math tools, a 50% larger graphing area, USB-C charging, and a simplified keypad with clearer commands. It includes smarter menus, built-in help via a yellow status bar, and comes in multiple color options (white, pink, mint, raspberry, silver, teal, lavender). Marketed as a distraction-free learning tool designed for durability from middle school through college, it emphasizes focus for classroom and high-stakes exam use.
Hacker News comments focused on the $160 price, with many expressing surprise it wasn't higher and questioning its value compared to modern hardware. Key points include the inclusion of a basic Python environment sparking interest and concerns about exam implications. There was significant criticism of TI's artificial product differentiation, particularly the lack of Computer Algebra System (CAS) despite being standard for years. Many comments debated the relevance of physical calculators in the software era, suggesting Casio alternatives or software solutions offer better value. The discussion also highlighted frustration over TI's monopolistic pricing practices perceived as a "ripoff" in academia.
HN discussion
(107 points, 218 comments)
The post is an "Ask HN" thread titled "Who wants to be hired?" (May 2026), inviting individuals seeking employment to share their professional details using a structured format. Required fields include location, remote work preference, willingness to relocate, technologies, résumé/CV link, and email. The thread explicitly prohibits agencies, recruiters, and job boards, emphasizing that only personal job seekers should participate. Readers are directed to contact applicants via email for opportunities, and two external job-search resources (nthesis.ai and wantstobehired.com) are suggested.
The top comments showcase a diverse pool of candidates across global locations (e.g., US, Europe, Asia), with most prioritizing remote work or hybrid arrangements. Common themes include backend/backend-adjacent expertise (Python, Go, Node.js), AI/ML specialization (PyTorch, LangChain, agentic workflows), and experience with cloud technologies (AWS, GCP, Azure). Applicants range from early-career graduates to senior engineers with 20+ years of experience, often highlighting open-source contributions, product ownership, or significant technical achievements (e.g., optimizing AWS costs by $1M+, reducing app crashes). Some express frustration with lengthy interview processes or Kubernetes overengineering, while others emphasize pragmatic, "grug-brain" approaches to development.
HN discussion
(236 points, 65 comments)
Residents of Dunwoody, Georgia, discovered that Flock employees had been accessing surveillance cameras, including those in sensitive locations like a children's gymnastics room, a playground, a school, a Jewish community center, and a pool, to demonstrate their technology to police departments. Flock confirmed the access was part of an authorized "demo partner program" with the city but denied allegations of spying, claiming it was for testing and debugging with the city's permission. The company defended its transparency, stating it creates access logs available via public records requests, and has since agreed to stop using sensitive location cameras for demos, restricting them to more public areas like parking lots.
The top HN comments focus on the broader implications of the surveillance system and Flock's practices. Many commenters question why Flock uses live cameras for demos instead of a dedicated environment, noting that this raises significant privacy concerns. Others highlight the irony of "saving children" rhetoric being used to justify the very surveillance system that exposes them to potential abuse. The discussion also criticizes Flock for removing barriers to access, arguing that enabling widespread, real-time surveillance is fundamentally different from targeted law enforcement access and constitutes a dangerous "privacy nightmare."
HN discussion
(109 points, 173 comments)
Peter Roberts, an immigration attorney specializing in YC and startup cases, hosted a 6-hour AMA session to answer general immigration questions. He emphasized he cannot provide case-specific legal advice due to lack of full facts and requested factual discussions. Roberts referenced previous similar AMAs for context.
Key questions focused on visa specifics, including typical timelines and costs for H-1B, EB-3, and other categories; paths from zero to citizenship for non-EU Balkans residents; and the impact of recent policies like the $100k H-1B fee. Concerns centered on increased denial rates, processing delays (especially for EB visas), and difficulties with permanent residency applications under the current administration. Practical issues addressed included risks around F-1 to H-1B status changes during travel, ACA health plan eligibility on B-2 visas, and rules regarding employment-based green card job duration. A personal testimonial highlighted Roberts' effectiveness in securing critical work authorization for startup success.
HN discussion
(152 points, 119 comments)
The article describes how the author's credit card was stolen through a brute force attack despite PCI DSS compliance. Attackers obtained masked card numbers and expiration dates from breached accounts, then systematically tested combinations at a rate of 6 requests per second via proxy-spoofed IPs. They exploited merchants exempt from 3D Secure authentication, withdrawing funds to e-wallets. PCI DSS standards allow displaying the first 6/8 digits (BIN) and last 4 digits of a PAN, which inadvertently aided attackers. The author received a full refund via chargeback but criticized payment systems for inadequate safeguards, noting that some gateways process transactions with minimal verification and that merchants dismissed vulnerabilities as compliant with standards.
Hacker News comments emphasized practical consumer protections, such as using dedicated virtual cards with low balances for online purchases to limit exposure. Users highlighted superior fraud protection with credit cards versus debit cards, where liability falls on consumers. Many criticized the credit card system's outdated design (e.g., static card numbers, optional 3D Secure in the US), noting that dynamic keys (e.g., Amex's rotating CVV) or blockchain payments (Lightning Network) could enhance security. Others pointed out flaws in the settlement process, where banks authorize transactions without robust authentication, shifting liability to merchants. Digital wallets were noted as complicating card cancellations, requiring manual intervention to disable all linked services.
Generated with hn-summaries