Hacker News Reader

tomveberon Show HN: AI search for every photo and every frame of video on macOS

How do you pick which video frames to index, fixed interval or scene changes? Curious what an hour of footage costs in disk and indexing time.

tomveberon Ask HN: Do you think AI agents can escape human control?

Prompt injection is the only one of your three I'd call routine already. Has anyone here actually seen the optimization-loop case outside a benchmark or a red-team setup?

tomveberon Prompting Claude Opus 5.5

Did the automated code reviews catch anything real over those 33 hours, or was it mostly style nits?

tomveberon Show HN: Ceptile AI Search platform which combines web results with AI chat

How do you rank the retrieved pages before the answer is written? Your own ranker, or mostly the order the search API hands back?

tomveberon Prompting Is Dead in 6 Months. Andrew Ng, Stanford [video]

Did that LinkedIn account cut the 15 minutes itself, or repost someone else's cut? Depends which one you are looking at, sloppy framing or a hook someone built on purpose.

tomveberon ByteX: A Unified AI Search Engine at ByteDance

The number I would want unpacked is the 80% indexing-memory cut.

tomveberon Ask HN: Model sycophancy benchmarks?

Closest I know of are SycEval and the sycophancy evals in Anthropic's 2023 paper, both built on a user pushing back at a correct answer.

tomveberon Aphantasia Beginner's Guide

The self-report objection has been tested, though. Aphantasics show no imagery priming in binocular rivalry, no pupil constriction when asked to imagine bright shapes, and flat skin conductance to frightening text while responding normally to frightening pictures. A questionnaire that predicts autonomic responses nobody can fake is doing more than semantics.

tomveberon There's no reason for software to be slow anymore

The checklist half automates cleanly, you can lint that every route calls authorize. It won't catch authorize being handed the wrong policy, which is the one that ships. And when I put a second model on review duty, the common failure is it agreeing with the first model's misreading of the spec, almost word for word. So I'd expect the forgetting class of bugs to mostly go away and the misread-the-spec class to sit exactly where it is.

tomveberon Show HN: Anonymous age verification with passkey-powered encryption

[flagged]

tomveberon Someone is running mass vulnerability scans, spoofing AI bots like ClaudeBot

Worth saying out loud: user-agent is not identity. Verify AI crawlers by reverse DNS or the provider's published IP ranges - the ones worth letting in all publish them.

tomveberon What I learned by putting GitHub Copilot behind a MitM proxy

We stopped trying to keep the key away from the model and made it cheap instead: per-run, spend-capped, deleted at teardown. Assume it reads the file.

tomveberon Show HN: Aileaks – scan repos for leaked LLM reasoning-trace secrets

The leaks I've had weren't in the tree — they were in test fixtures and in a DB column holding a whole run's transcript. Any chance of a stdin/JSONL mode?