Hacker News
1Tell HN: Anthropic no longer allowing Claude Code subscriptions to use OpenClaw
- Starting April 4, Anthropic is restricting Claude subscription usage on third-party tools like OpenClaw, requiring separate pay-as-you-go billing for these services. Users can still access core Claude products with their subscription and receive a one-time credit plus usage bundle discounts to ease the transition.
Subscription model economics
Users debate how subscription services work - they oversell capacity with light users subsidizing power users. OpenClaw's autonomous 24/7 usage breaks this model, forcing Anthropic to either raise prices for everyone or ban heavy automated usage patterns.Capacity constraints vs pricing
Discussion centers on whether this is about financial impact or GPU capacity limits. Many believe Anthropic is capacity-constrained and prioritizing enterprise customers over individual power users, rather than purely cost-driven decisions.Platform limitations and alternatives
Users express frustration with arbitrary rule changes and consider switching to alternatives like OpenAI, Chinese models, or local solutions. Many question building on Claude as a platform given unpredictable policy changes.
2Author of "Careless People" banned from saying anything negative about Meta
- Meta silenced author Sarah Wynn-Williams after she wrote "Careless People," a book alleging sexual harassment and censorship at the tech giant. Despite legal threats and attempts to suppress her work, her publisher argues Meta's actions actually validate her claims about the company's problematic culture and practices.
Powerful people receiving unethical passes
Discussion centers on how society enables wealthy executives' inappropriate behavior, with specific focus on Sheryl Sandberg's alleged misconduct. Users debate whether we give rich people different moral standards than we would accept for children or teach others.Non-disparagement clause enforcement
Users debate the ethics and legality of Meta enforcing a 9-year-old severance agreement to silence the author. Discussion covers whether taking money while planning to violate contracts is principled, and concerns about corporate power to suppress speech.Author credibility and complicity
Mixed views on whether the whistleblowing author should be considered heroic or complicit. Some argue she was deeply embedded in Meta's culture and only spoke out after being excluded, while others defend the value of insider accounts regardless of motivation.
3F-15E jet shot down over Iran
Military effectiveness and Iran's defensive capabilities
Users debate whether the F-15 shootdown indicates Iran's air defenses remain effective despite weeks of bombing, comparing it to Gulf War losses and discussing modern SAM technology improvements versus stealth capabilities.Treatment of captured pilots and war terminology
Discussion centers on whether captured airmen are POWs or hostages, SERE training effectiveness, and concerns about pilot treatment given Iran's human rights record and the undeclared nature of the conflict.Content moderation and bias on Hacker News
Users question why this military story stays up while posts about Palestinian casualties get flagged, debating HN's content guidelines and perceived bias in what qualifies as "tech" versus "politics."
4The FAA’s flight restriction for drones is an attempt to criminalize filming ICE
- The Trump administration implemented a nationwide flight restriction preventing drone operators from flying within half a mile of ICE or CBP vehicles, effectively restricting First Amendment rights to record law enforcement. This 21-month "temporary" restriction violates constitutional rights and lacks proper justification, as EFF argues in demanding its removal.
Enforcement concerns and legal loopholes
Discussants worry the rule creates impossible compliance scenarios where drone operators can unknowingly violate laws when unmarked ICE vehicles approach. Key concerns include mens rea requirements, selective enforcement, and the rule's true purpose being to suppress viral footage rather than safety.Constitutional and civil rights violations
Multiple commenters argue this represents government overreach that violates First Amendment rights to record law enforcement, creates a "dual state" with arbitrary enforcement, and uses emergency powers to justify authoritarian control over public oversight.Technical workarounds and resistance strategies
Technical users discuss building custom drones without tracking capabilities, using privacy-focused firmware, operating anonymously, and creating offshore platforms to share footage - essentially an arms race between surveillance and counter-surveillance.
5Embarrassingly simple self-distillation improves code generation
- Researchers developed Simple Self-Distillation (SSD), a method that improves large language model code generation by fine-tuning models on their own outputs without external verifiers or teacher models. SSD improved Qwen3-30B-Instruct's performance from 42.4% to 55.3% on LiveCodeBench by reshaping token distributions to balance precision and exploration in different contexts.
Fork and lock positions in code generation
Discussion centers on how LLMs need to switch between "exploration" mode at decision points (forks) and "precision" mode for syntactic correctness (locks). The Simple Self-Distillation technique helps models better navigate both modes, improving token ranking.Emergent properties and understanding of LLMs
Participants debate whether discovering new LLM capabilities is surprising, comparing AI research to studying human brains. A psychiatry resident notes we may understand LLMs better than humans in some ways, while others argue such discoveries are expected given LLMs' black-box nature.Self-training effectiveness and model collapse concerns
Users question why training models on their own outputs was thought to degrade quality, when this paper shows improvements. Discussion touches on the contradiction with previous "model collapse" research and whether self-distillation represents a breakthrough.
6iNaturalist
- This platform enables users to record wildlife encounters, create species life lists, contribute scientific data, get expert identification help, participate in citizen science projects, learn about nature through community interaction, and organize bioblitz events to discover local biodiversity.
iNaturalist API and apps
Users praise the iNaturalist API for being accessible without authentication and having open CORS headers, making it excellent for demos. Multiple related apps are discussed including Seek, Merlin Bird ID, and Flora Incognita for species identification.Privacy concerns with location data
Discussion about how iNaturalist can inadvertently reveal users' home addresses through observation patterns, especially affecting non-technical users who may not realize their privacy implications or know about geoprivacy settings.Community impact and citizen science
Users share how iNaturalist transformed their relationship with nature, helping them learn local flora and fauna, contribute to scientific research, and engage with wildlife communities during the pandemic and beyond.
7Oracle files H-1B visa petitions amid mass layoffs
- Oracle filed over 3,100 H-1B visa petitions in recent fiscal years while simultaneously laying off thousands of American workers this week. The Austin-based software company filed 2,690 visas for fiscal 2025 and 436 for fiscal 2026, raising questions about replacing US workers with cheaper foreign labor during organizational restructuring.
H-1B visa abuse and job posting manipulation
Discussion centers on companies allegedly posting H-1B job openings in obscure locations to avoid qualified US applicants, then claiming no domestic workers are available. Users share examples and debate enforcement effectiveness.Oracle layoffs versus H-1B filings timing and scope
Debate over whether Oracle's H-1B applications preceded recent layoffs, with conflicting claims about layoff numbers in US vs India. Users question the appropriateness of hiring foreign workers while conducting layoffs.$100k H-1B fee impact and loopholes
Users analyze the new $100,000 H-1B fee's actual effect, noting it only applies to new overseas hires, not status changes for existing US residents. Discussion reveals limited enforcement with only 85 applications paying the fee.
8OpenClaw privilege escalation vulnerability
- The content contains only two URLs - one linking to a Reddit post about OpenClaw security issues and another linking to an archived version of the same post. There is no actual readable content to summarize beyond the URL titles suggesting OpenClaw users may have been compromised.
Severity and scope of the vulnerability
OpenClaw creator clarifies the CVE was a privilege escalation bug requiring existing gateway access, not a direct remote exploit. However, discussion reveals 135k+ publicly exposed instances with 63% running zero authentication, creating significant security risks despite the technical requirements.Security architecture and sandboxing approaches
Users discuss various isolation strategies including running OpenClaw as limited user accounts, using VMs, and kernel-level sandboxing. Many emphasize the fundamental security challenge of giving AI agents system access and the need for proper containment mechanisms.OpenClaw's security track record and development practices
Community highlights OpenClaw's concerning history of 1.8 CVEs per day since launch and questions the sustainability of AI-generated code. Discussion includes skepticism about corporate partnerships and the broader implications of "vibe coding" for software security.
9Delve removed from Y Combinator
- This is a Y Combinator website 404 error page indicating that the requested file was not found. The page provides a link to return to the homepage and offers a support contact email at software@ycombinator.com for assistance.
Beyond license violations - serious fraud allegations
Many commenters clarify that Delve's ejection wasn't primarily about Apache 2.0 license violations, but rather allegations of fraudulent compliance audits including pre-written conclusions, rubber-stamping non-compliant customers, and fake SOC2 reports that exposed clients to criminal liability.YC's trust-based community model at risk
Discussion centers on how YC's model relies on batch companies trusting each other, and Delve's alleged fraud fundamentally breaks this trust. Comments note this threatens the entire YC ecosystem where early companies depend on each other for critical services.Forbes 30 Under 30 correlation with fraud
Multiple commenters observe a pattern of Forbes 30 Under 30 recipients being involved in major fraud cases (Holmes, SBF, Shkreli), questioning whether the personality traits that get someone on such lists correlate with willingness to engage in fraudulent behavior.
10Show HN: A game where you build a GPU
- The author noticed a lack of available resources about GPU architecture and decided to create content to address this gap.
User experience and interface issues
Users reported UI problems including confusing grid lines that looked like wires, timer issues in minigames, misaligned connections, and requests for hint systems and skip options for advanced users.Technical accuracy concerns
Discussion about capacitor implementation being unrealistic (having built-in enable gates), truth table quiz timing issues, and suggestions to improve educational sequencing by teaching concepts before testing.Learning accessibility
Many users struggled with basic concepts, requesting expanded acronyms, introductory levels, and better explanations of fundamentals like 3-state logic and transistor physics.