AI solves ancient cipher & Google ad reviews questioned - Hacker News (Sep 14, 2026)

Please support this podcast by checking out our sponsors: - SurveyMonkey, Using AI to surface insights faster and reduce manual analysis time - https://get.surveymonkey.com/tad - Consensus: AI for Research. Get a free month - https://get.consensus.app/automated_daily - Discover the Future of AI Audio with ElevenLabs - https://try.elevenlabs.io/tad Support The Automated Daily directly: Buy me a coffee: https://buymeacoffee.com/theautomateddaily Today's topics: AI solves ancient cipher - Vals AI says Claude Fable 5.1 solved Sir Thomas Urquhart’s 370-year-old Cyphral Distich by treating the numbers as references back into the book. The result, a royalist prayer for King Charles II, highlights how AI models may now assist with obscure historical ciphers and archival research. Google ad reviews questioned - A writer reported deceptive iPhone-style ads in YouTube, and Google reportedly kept approving them even after multiple complaints. The twist is that Google’s own Gemini model flagged the ad as misleading immediately, raising questions about ad safety, moderation quality, and platform incentives. Signal moves beyond numbers - Signal appears to be actively building registration and login without a phone number, based on new Android commits and discussion around identity and spam control. It matters because numberless accounts could improve privacy while forcing Signal to balance abuse prevention and account recovery. Julia 1.13 boosts speed - Julia 1.13 has arrived with faster startup, lower package precompilation time, and a more polished REPL. The release matters for developers because it improves day-to-day usability while also making scheduling, interrupts, and tooling more reliable. JPEG XL web fight - A new essay argues JPEG XL is technically strong but still a poor fit for browsers compared with AVIF and WebP. The debate matters because web image standards affect performance, compatibility, and how much complexity browsers should absorb for niche gains. Automattic leadership snaps back - Automattic says Matt Mullenweg is back as chairman and CEO after a confusing board move that briefly put him on leave. The reversal matters because Automattic sits at the center of a major WordPress ecosystem, and leadership instability could ripple outward. - Claude Fable 5.1 Solves a 370-Year-Old Cipher - Why Google Keeps Approving Deceptive Ads - Signal Advances Numberless Account Registration - Apple Developer Dimensional Drawings Resource - Julia 1.13 Release Brings Faster Startup, Better REPL, and GC Improvements - Mullenweg Returns as Automattic CEO After Board Ouster Attempt - Why JPEG XL May Not Be Right for the Web - Nike Drops Out of the S&P 100 After Massive Market Cap Loss - Open-Source AI and Open Models Reading List Episode Transcript AI solves ancient cipher We’ll start with that AI puzzle story, because it’s the one most likely to make historians and model builders stop and stare. Vals AI reports that Claude Fable 5.1 solved Sir Thomas Urquhart’s Cyphral Distich, a 17th-century cipher that had resisted earlier attempts. The key insight was apparently simple in hindsight: the numbers pointed back into the book itself, not to some outside code table. The decoded message turned out to be a royalist prayer supporting King Charles the Second. If the report holds up, the real significance is not just one solved riddle. It’s that frontier models may now be genuinely useful for slow, obscure research problems that used to require a lot of patient human effort. Google ad reviews questioned Staying with AI, but in a much less romantic setting, there’s a pointed critique of Google’s ad moderation. The complaint is that deceptive ads designed to look like iPhone system alerts are still being approved inside YouTube, even after users report them. What makes the story sting is that the author says Google’s own Gemini model immediately recognized the ad as misleading and something that should be rejected. So the gap here is not whether the problem is detectable. It seems to be whether Google’s enforcement process is actually acting on it. That matters because ad scams scale fast, and trust in a platform drops quickly when the fake stuff is easier for AI to spot than for the company to remove. Signal moves beyond numbers On the privacy front, Signal appears to be moving closer to accounts that don’t depend on phone numbers. The discussion centers on recent Android commits that suggest numberless registration and login are actively being built. That’s a meaningful shift for a service that has long tied identity to a phone number, even while trying to minimize what that number reveals. The open questions are exactly the ones you’d expect: how do you preserve privacy, how do you recover accounts, and how do you stop spammers from exploiting easier signup paths? The big takeaway is that this no longer looks like a vague idea. It looks like work in progress, and potentially one of Signal’s biggest account-model changes in ye