OpenSSL: Buffer overflow in EVP_EncodeBlock() (CVE-2026-121481, CVSS 9.8)
A commenter notes that EVP_EncodeBlock malfunctions if the input is larger than ~1.5GB and that the API is poorly designed as it does not validate the output buffer length.
30 stories
Get a notification every evening
You hear about the digest when it is ready. Turn it off any time.
Notifications are on
You will hear about each evening’s digest when it is ready.
To get notifications on iPhone or iPad, add this site to your Home Screen first: tap Share, then Add to Home Screen. Open it from there to turn notifications on.
Notifications are blocked in this browser. Allow them in site settings and reload.
New notification sign-ups are closed for now. You can follow along with RSS.
Notifications could not be turned on. Check your connection and try again.
Notifications could not be turned off and are still on. Check your connection and try again.
The day’s top discussions
A commenter notes that EVP_EncodeBlock malfunctions if the input is larger than ~1.5GB and that the API is poorly designed as it does not validate the output buffer length.
A commenter describes reverse engineering their CCTV camera's cloud connection plugin with Gemini in about 2 hours.
A commenter suggests users write to their state attorneys general to investigate Google's pressure on OEMs from an antitrust perspective.
A commenter notes that the only improvement the model made was to shorten its reasoning traces to avoid hitting the token limit.
A commenter says ID.me requires a smartphone to sign up, causing problems for someone unable to cash out Treasury Direct bonds.
A commenter (Quad9's CTO) states there is full support for Quad9's position at IETF and a general sense of dread about DNS developments.
A commenter notes that interaction nets are a fundamentally different programming paradigm and suggests the discussion should focus on the topic.
A commenter is excited about the ZTS overhead reduction and stream error handling improvements in PHP 8.6.
Commenters discuss that Gemini 4 Neon is fast and competitive, but access to the model is difficult.
Commenters discuss the reporting process of the java4shell bug to Oracle and that simultaneous disclosures are common.
A commenter notes that Claude's second result proves the upper limit for black hole mass, making it a consistent quantum gravity theory.
A commenter notes that the INSERT ... ON CONFLICT DO SELECT feature in PostgreSQL 19 simplifies the 'get or create' pattern.
A commenter hopes that some features from the qemu-wearables fork made for Cuttlefish will be upstreamed.
One commenter agrees with the author, noting that five years ago the HN consensus was that moving away from code atrophies skills; another criticizes the industry's blind acceptance of AI promises.
One commenter says Epic fails at AI coding basics, official tools are nearly useless; another shares discarding a Claude-made Godot project for not feeling like his own art.
One commenter questions why macros aren't used for deterministic lookup tables; another pays tribute to the late Kevin Lang.
One commenter appreciates the specificity of domain knowledge integration in early neural networks; another thanks the article for demystifying weight-space programming as methodical reasoning.
One commenter praises the GPL license, calling many Show HN projects mercenary; another hails natural search as mundane yet revolutionary, impressed by the deterministic approach.
A commenter says Google isn't clear on how to pay for Genie 4 and that version may differ due to being in the UK.
A commenter says LLMs are useful for remembering all details on tedious topics, making auditing a perfect use case.
A user says they've been using JJ almost exclusively for two years and looks forward to JJ-native forges.
A commenter says weight editing is not the only option to break loops; repetition penalties sometimes work.
A commenter says the Fable 5.2 and Opus 5.5 combination has led to the most robust software results so far.
A smaller, more technical community
A commenter asks whether the previously unknown binder.c IPC mechanism is a counterpart of iOS's xpc API.
One commenter supports the project but warns against limiting rewrites to Rust; another questions the inclusion of Go and Python as memory-safe alternatives.
The week’s most-starred new repositories