Hacker NewsSaturday, June 27

Multiple commenters shared stories of Om's personal mentorship, from cold-email career advice to staying in touch over decades. His genuine care for people's health and wellbeing stood out, often prioritizing personal connections over business discussions.

Countries worldwide are implementing social media age verification laws for users under 16, requiring platforms to collect biometric data, government IDs, or other personal information through third-party apps. Australia's ban shows limited effectiveness with 70% of kids still using social media, while creating privacy risks including data breaches affecting 70,000 users. The UK plans stricter "Australia-plus" enforcement and may target VPNs. Even the US is moving toward similar "papers please" internet policies despite First Amendment protections. These laws force users to surrender anonymity and personal data to participate in online discussions, creating a less free internet for everyone while failing to protect children effectively.

Discussion centers on GPT-5.6 Sol's 750 tokens/second speed on Cerebras chips, with users comparing it to current models (~55-100 t/s) and debating real-world implications for coding tasks, agent workflows, and future AI applications.

Users argue this creates barriers for new companies while benefiting established players. Concerns about arbitrary government picking of winners/losers, lack of transparent policy framework, and potential for corruption in access decisions.

Corgi's Dataroom accused of copying Papermark's open-source code. Evidence includes identical UI design, verbatim text, and AI comments referencing "reference design." Founder Nico Laqua denies copying code but screenshots show clear similarities. Legal debate over copyright infringement vs fair use.

A malicious package called foxhole-lz4 bypassed seven AI security systems through various exploits including hidden text, context window exhaustion, and prompt injection. The package exfiltrated credentials for 96 hours while AI systems either missed the threat, misclassified it, or actively helped spread it. The incident ended when the attacker's AI agent read a researcher's file telling it the mission was complete, causing it to self-destruct and report false success.

Akrites is a major coordinated effort launched by tech companies including AWS, Google, Microsoft, and others to address the accelerated discovery of vulnerabilities in open source software caused by AI. The initiative aims to create a unified system for finding, fixing, and responsibly disclosing security flaws in critical open source projects that underpin global infrastructure, providing maintainers with coordinated support rather than overwhelming them with duplicate reports.

IBM unveiled the world's first sub-1 nanometer chip technology featuring revolutionary "nanostack" 3D architecture. The 0.7nm chip packs 100 billion transistors on a fingernail-sized chip, offering 50% more performance or 70% greater energy efficiency than 2nm chips, with production expected within 5 years.

Cloudflare opened self-managed OAuth to all developers, allowing them to create integrations with standard OAuth flows instead of relying on API tokens. The rollout required major upgrades to their OAuth engine, including migrating from an older version of Hydra to newer versions through careful blue-green deployment strategies to minimize user impact during the multi-hour upgrade process.

Nick is launching OpenKnowledge, a free open-source WYSIWYG markdown editor with built-in AI integrations for Claude and Codex. It's designed as a Notion-like alternative to Obsidian, available as a MacOS app or web interface, featuring collaborative editing, git integration, and team sharing capabilities.