frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Ask HN: How are you preventing LLM hallucinations in production systems?

1•kundan_s__r•1h ago
Hi HN,

For those running LLMs in real production environments (especially agentic or tool-using systems): what’s actually worked for you to prevent confident but incorrect outputs?

Prompt engineering and basic filters help, but we’ve still seen cases where responses look fluent, structured, and reasonable — yet violate business rules, domain boundaries, or downstream assumptions.

I’m curious:

Do you rely on strict schemas or typed outputs?

Secondary validation models or rule engines?

Human-in-the-loop for certain classes of actions?

Hard constraints before execution (e.g., allow/deny lists)?

What approaches failed for you, and what held up under scale and real user behavior?

Interested in practical lessons and post-mortems rather than theory.

Comments

al_borland•1h ago
I’ve just been ignoring my boss every time he says something about how we should leverage AI. What we’re building doesn’t need it and can’t tolerate hallucinations. They just want to be able to brag up the chain that AI is being used, which is the wrong reason to use it.

If I was forced to use it, I’d probably be writing pretty extensive guardrails (outside of the AI) to make sure it isn’t going off the rails and the results make sense. I’m doing that anyway with all user input, so I guess I’d be treating all LLM generated text as user input and assuming it’s unreliable.

kundan_s__r•1h ago
That’s a very sane stance. Treating LLM output as untrusted input is probably the correct default when correctness matters.

The worst failures I’ve seen happen when teams half-trust the model — enough to automate, but still needing heavy guardrails. Putting the checks outside the model keeps the system understandable and deterministic.

Ignoring AI unless it can be safely boxed isn’t anti-AI — it’s good engineering.

stephenr•14m ago
I've found that I can use a very similar approach to the one I've used when handling the risks associated with blockchain, cryptocurrencies, "web scale" infrastructure, and of course the chupacabra.

Bottom-up programming as the root of LLM dev skepticism

https://www.klio.org/theory-of-llm-dev-skepticism/
1•mkozlows•2m ago•0 comments

Show HN: Neutriva – A personalized health and wellness tracking assistant

https://neutriva.com/en/wellness-assistant
1•NoraWW•3m ago•0 comments

Medical Groups Will Try to Block Childhood Vaccine Recommendations

https://www.nytimes.com/2026/01/13/health/vaccine-schedule-children-kennedy.html
2•doener•12m ago•0 comments

Terry Tao: "LLMs Are Simpler Than You Think – The Real Mystery Is Why They Work" [video]

https://www.youtube.com/watch?v=ukpCHo5v-Gc
2•gmays•12m ago•0 comments

Starlink Users in Iran Get Free Internet Access, Nonprofit Says

https://www.nytimes.com/2026/01/13/technology/iran-starlink-elon-musk.html
2•doener•14m ago•0 comments

One Simple Arrow Changed Automobiles Forever [video]

https://www.wsj.com/video/series/on-the-news/how-one-simple-arrow-changed-automobiles-forever/C33...
1•fortran77•16m ago•0 comments

Qualcomm's RISC-Ventana Fusion

https://thechipletter.substack.com/p/qualcomms-risc-ventana-fusion
1•chmaynard•17m ago•0 comments

What's Ahead: Alien Processes, Domains, and Data Models

https://practicaldatamodeling.substack.com/p/whats-ahead-alien-processes-domains
1•gmays•18m ago•0 comments

Why IRC is better than Real Life

https://everything2.com/node/e2node/Why%20IRC%20is%20better%20than%20Real%20Life
2•jskherman•20m ago•0 comments

A new generation of Chinese companies is expanding around the world

https://www.economist.com/business/2026/01/13/a-new-generation-of-chinese-companies-is-expanding-...
3•petethomas•22m ago•0 comments

Southern New Zealand hospitals experienced major IT outage

https://www.rnz.co.nz/news/national/584026/public-service-association-says-southern-hospitals-exp...
3•billybuckwheat•22m ago•0 comments

What will enshittification of LLMs look like?

1•scoofy•23m ago•2 comments

An Updated Dentist Office Software Story

https://avc.xyz/an-updated-dentist-office-software-story
1•turadg•24m ago•1 comments

BioNTech Provides Strategic Business Update and Outlines 2026

https://investors.biontech.de/news-releases/news-release-details/biontech-provides-strategic-busi...
2•doener•26m ago•0 comments

How to Store the Web in S3

https://exa.ai/blog/exa-d
1•willbryk•27m ago•1 comments

Signal creator Moxie Marlinspike wants to do for AI what he did for messaging

https://arstechnica.com/security/2026/01/signal-creator-moxie-marlinspike-wants-to-do-for-ai-what...
3•abolishme•29m ago•0 comments

Lawsuit: DHS wants "unlimited subpoena authority" to unmask ICE critics

https://arstechnica.com/tech-policy/2026/01/instagram-user-fights-dhs-for-the-right-to-post-ice-s...
4•duxup•31m ago•0 comments

OpenAI buys tiny health records startup Torch for, reportedly, $100M

https://techcrunch.com/2026/01/12/openai-buys-tiny-health-records-startup-torch-for-reportedly-100m/
1•nsoonhui•31m ago•0 comments

China's durian craze has turned this tropical fruit into a tool of diplomacy

https://theconversation.com/chinas-durian-craze-has-turned-this-tropical-fruit-into-a-tool-of-dip...
2•PaulHoule•32m ago•0 comments

Our Slapdash Cultural Change

https://www.overcomingbias.com/p/our-slapdash-cultural-change
1•paulpauper•33m ago•0 comments

MoneyRank – a daily 60-second game that scores your financial risk instincts

https://moneyrank.onrender.com/
1•abbster52•34m ago•1 comments

Vanderbilt University Plans New Campus in San Francisco

https://www.wsj.com/us-news/education/vanderbilt-san-francisco-cca-california-college-arts-expans...
1•noleary•34m ago•0 comments

Toyota remained top automaker by sales for 6th straight year in 2025

https://asia.nikkei.com/business/automobiles/toyota-remained-top-automaker-by-sales-for-6th-year-...
1•breve•35m ago•0 comments

Device that may be tied to "Havana Syndrome" obtained by U.S. government

https://www.cbsnews.com/news/device-havana-syndrome-obtained-by-u-s-government/
5•mhb•35m ago•3 comments

Why China Is Suddenly Obsessed with American Poverty

https://www.nytimes.com/2026/01/13/business/china-american-poverty.html
5•xnhbx•38m ago•3 comments

More Young Americans Are Unfit to Serve, a New Study Finds. Here's Why

https://www.military.com/daily-news/2022/09/28/new-pentagon-study-shows-77-of-young-americans-are...
3•paulpauper•39m ago•3 comments

Ask HN: Preserving knowledge long-term without a central authority

1•SERSI-S•40m ago•0 comments

Claude Coworks

https://thezvi.substack.com/p/claude-coworks
1•paulpauper•40m ago•0 comments

Former NYC Mayor Eric Adams accused of $2.5M rug pull as NYC Token crashes 80%

https://www.theverge.com/news/861269/former-nyc-mayor-eric-adams-accused-of-2-5-million-crypto-ru...
3•beeandapenguin•41m ago•1 comments

Show HN: Demo of Rust Lettre crate for sending email using SMTP

1•jph•42m ago•1 comments