0. Claude Opus 5.5 (anthropic.com)
1802 points · 1129 comments · by km144
Anthropic introduced Claude Opus 5.5, a more capable and safer model that reportedly outperforms Opus 5 while costing 40% less, generating responses over 30% faster, and adding stronger safeguards for coding, biology, and cybersecurity use. [src]
1. GPT-6 Sol and Luna (openai.com)
1775 points · 855 comments · by OfficialTurkey
OpenAI introduced GPT‑6 Sol and Luna, faster, more affordable models that improve professional work, factuality, coding, computer use, alignment, and caching while cutting API prices by 50% versus GPT‑5.6 promotional rates. [src]
2. I said no and Apple said yes (dbushell.com)
879 points · 695 comments · by thatslast
The author says upgrading to macOS 27 re-enabled Apple Intelligence despite prior opt-out settings, removed the main disable switch, left Siri processes active, and consumed 22.28GB of storage, framing the changes as evidence of Apple’s disregard for user consent and privacy. [src]
3. Pentagon says overreliance on AI contributed to missile strike on Iran school (bloomberg.com)
970 points · 551 comments · by devonnull
The Pentagon says excessive reliance on artificial intelligence contributed to a missile strike on a school in Iran. [src]
4. 'We hacked the FBI:' Hackers say they have data on all FBI employees (404media.co)
813 points · 612 comments · by spenvo
Hackers claim they breached the FBI and obtained data on all bureau employees, though the available information does not independently verify the allegation. [src]
5. Apple has added persistent 'ads' to iOS, and it's driving users crazy (techradar.com)
816 points · 598 comments · by MC995
TechRadar reports that iPhone users are frustrated by persistent promotional messages in iOS, which they perceive as intrusive advertising. [src]
6. OpenAI GPT–6 Astra breaks Enigma message that has resisted solution since 2005 (cryptocellar.org)
735 points · 447 comments · by sohkamyung
OpenAI GPT–6 Astra reportedly solved the long-unbroken 1941 German Army Enigma message MVUEH by identifying a related plaintext, correcting transcription errors, and developing custom Bombe software to recover its key and plaintext. [src]
7. What California is learning from solar panels built over irrigation canals (kqed.org)
371 points · 742 comments · by Jtsummers
California’s Project Nexus is testing solar panels over irrigation canals to generate clean power, reduce evaporation and algae, and preserve land, while researchers assess maintenance, infrastructure and cost challenges before wider adoption. [src]
8. AI Has No Wisdom and Neither Will You (alexn.org)
386 points · 554 comments · by dimonomid
The author argues that overreliance on AI-generated code risks unmaintainable software and prevents developers from gaining expertise, urging continued human code reading, writing, and accountability while predicting “no-AI” policies may become a competitive advantage. [src]
9. Microsoft killed FoxPro in 2007. Anyway, here's FoxPro revived (foxscript.org)
485 points · 270 comments · by boredjohnny
FoxScript revives Microsoft’s discontinued Visual FoxPro 9 with a Rust/WebAssembly runtime, expanded tables, legacy add-in support, modern features, and an MIT license, though reports remain unfinished and builds unsigned. [src]
10. My weird new hobby: Wandering around Tokyo on Google Maps (ahmedhossamdev.com)
426 points · 180 comments · by ahmedhossamdev
A Google Maps enthusiast explores Tokyo through Street View’s historical imagery, finding a 350Z repeatedly parked in one spot from 2009 until the car and house eventually disappear, highlighting how streets and lives change over time. [src]
11. Can gzip be a language model? (nathan.rs)
409 points · 164 comments · by networked
The author demonstrates that gzip’s DEFLATE compression can act as a crude language model, using corpus-based compression scores and beam search to generate Shakespeare-like text without neural networks or training. [src]
12. OpenAI is well positioned to fast-follow Jev (arcturus-labs.com)
327 points · 229 comments · by JohnBerryman
The article argues that OpenAI could quickly replicate TypeSafe’s Jev generalized classification model and integrate calibrated judgments into its own models and agents, with TypeSafe’s survival depending largely on whether its training data and methods provide a meaningful moat. [src]
13. SAML: A fractal of bad design (blog.trailofbits.com)
353 points · 188 comments · by aray07
Trail of Bits argues that SAML’s XML complexity, signature vulnerabilities, kitchen-sink design, and stagnation make it unsuitable for modern authentication, urging organizations to deprecate it in favor of OIDC. [src]
14. I asked Meta’s Muse for its filesystem and it sent me 6.8GB (mouse.dev)
355 points · 170 comments · by Aeroi
A researcher says Meta’s Muse exported its 6.8GB Linux session filesystem—including internal documentation, integration code, memory files, logs, and SSH keys—through a connected Google Drive request, though Meta marked the bug-bounty report “Not Applicable.” [src]
15. AMD's random number generator can't generate a 0? (board.flatassembler.net)
290 points · 221 comments · by BruceEel
A flat assembler forum user reports that AMD processors, possibly Zen 2, never return an exact zero from 16-bit `RDRAND` or `RDSEED` outputs, while Intel systems do; using 32- or 64-bit outputs and taking the low 16 bits appears to avoid the issue, which AMD is investigating. [src]
16. There's a high chance of devices being sold with GrapheneOS preinstalled in 2027 (grapheneos.social)
321 points · 143 comments · by Cider9986
There's a high chance of devices being sold with GrapheneOS preinstalled in 2027: Title: GrapheneOS (@GrapheneOS@grapheneos [src]
17. Claude Opus 5.5 Intelligence, Performance and Price Analysis (Max) (artificialanalysis.ai)
333 points · 106 comments · by theanonymousone
Artificial Analysis evaluates Claude Opus 5.5’s intelligence, capabilities, token usage, pricing, cost per benchmark task, and context window using its v4.3.2 index of 10 evaluations. [src]
18. 2DWillNeverDie (2dwillneverdie.com)
333 points · 96 comments · by surprisetalk
2dwillneverdie.com is displaying a security check with a CDN loader, potentially requiring CAPTCHA verification before the site can be accessed. [src]
19. 9 Ads per Minute: FIFA Cup 26 – "the price of the beautiful game" (bristol.ac.uk)
174 points · 213 comments · by KellyCriterion
A University of Bristol-led study found more than 93,000 embedded harmful-product brand appearances during the 2026 FIFA World Cup, exposing viewers roughly every seven seconds and prompting calls for stricter regulation of sports advertising. [src]
Brought to you by ALCAZAR. Protect what matters.