Hallucinated facts, APIs and code

You can't tell. Humans are now hallucinating what is, or is not, AI.

FrustratedMonky · Nov 3, 2026 · comment in “Nvidia – open-source auto-ontology”

AI agents hallucinate prices: LLM grounding with live data

Dylan-Hayes · Nov 3, 2026 · story in “AI agents hallucinate prices: LLM grounding with live data”

The data isnt really there, in short, the depth is hallucinated.

burnoutdv · Nov 3, 2026 · comment in “Vathographs: Photographic depth maps as creative expression”

I agree, but also, I kind of feel like real judgement and understanding are a different thing from this "being good at thinking", and that in fact, this is the reason AIs hallucinate in the first place (A generated text can fit the pattern of a good response because its accurate, or because it's bs but has similar words and structure to other texts on the topic.

ERcel · Nov 2, 2026 · comment in “Astra can read and write weights to improve Qwen”

No to actually manipulate figures for me or fill something. I check every hallucination to actual documentation from the government.

kazinator · Nov 2, 2026 · comment in “Ask HN: Is anyone going to use AI only to file their taxes?”

If you mean 'an LLM' as in the owner/maintainer of the LLM/weights, then examples of this already exist. Whether it's DeepSeek/Gwen downplaying Uyghur forced labor camps [0] or Grok hallucinating a 'white genocide' in South Africa [1], it is clear that the owners of LLMs have political agendas that are expressed in the weights of popular LLMs.

exogenousdata · Nov 2, 2026 · comment in “If you love a website, back it up”

Prior to that, I've built an ai workflow engine and a platform for both minimizing hallucinations and working with documents that are far too large for existing context window inputs or outputs using a novel, patented process that I invented; and applied these tools to various contracted projects requiring analysis and and cross-checking of information and other forms of integrity across textual and video/audio corp…

poeticsilence · Nov 2, 2026 · comment in “Ask HN: Freelancer? Seeking freelancer? (November 2026)”

Prior to that, I've built an ai workflow engine and a platform for both minimizing hallucinations and working with documents that are far too large for existing context window inputs or outputs using a novel, patented process that I invented; and applied these tools to various contracted projects requiring analysis and and cross-checking of information and other forms of integrity across textual and video/audio corp…

poeticsilence · Nov 2, 2026 · comment in “Ask HN: Who wants to be hired? (November 2026)”

They probably asked Claude. Sounds like a normal LLM hallucination.

traceroute66 · Nov 2, 2026 · comment in “Anthropic reports $17B in Q3 revenue, commitments of $895B in revised S-1”

I no longer share them though. It’s the shit tier free or super low cost AI hallucination engines used by the masses that are the problem.

oefrha · Nov 2, 2026 · comment in “AI fan wikis riddled with errors are muddying the waters of games culture”

Bzzzt! Calculators do not hallucinate.

kazinator · Nov 2, 2026 · comment in “AI delivers 'zero financial payoff' to most companies, Apollo economist says”

In fact you can induce a sort of model collapse in a human by isolating them in a sensory-deprivation chamber, after which they will literally begin to hallucinate.

bitwize · Nov 1, 2026 · comment in “Patrick Stewart taught us what good men do with power”

I'd argue the improvement has come not so much from fixing hallucinations, but more in getting LLMs to successfully run for longer, which lets them more effectively fix their own hallucinations.

danpalmer · Nov 1, 2026 · comment in “So I guess everyone uses LLMs to think now”

> > i don't know how more people don't experience this, but staring at an llms output be confidently wrong eventually to get there in the process, or looking at the commands its running and see some really stupid things its doing drives me crazy to the point i will just stop it before it answers, i cannot handle something being that confidently wrong (even though it'll probably still get there in the end)

rolosa · Nov 1, 2026 · comment in “So I guess everyone uses LLMs to think now”

i don't know how more people don't experience this, but staring at an llms output be confidently wrong eventually to get there in the process, or looking at the commands its running and see some really stupid things its doing drives me crazy to the point i will just stop it before it answers, i cannot handle something being that confidently wrong (even though it'll probably still get there in the end)

CamperBob2 · Nov 1, 2026 · comment in “So I guess everyone uses LLMs to think now”

> i cannot handle something being that confidently wrong

lazyasciiart · Nov 1, 2026 · comment in “So I guess everyone uses LLMs to think now”

> don't know how more people don't experience this, but staring at an llms output be confidently wrong eventually to get there in the process, or looking at the commands its running and see some really stupid things its doing drives me crazy to the point i will just stop it before it answers, i cannot handle something being that confidently wrong (even though it'll probably still get there in the end)

tayo42 · Nov 1, 2026 · comment in “So I guess everyone uses LLMs to think now”

Good on them for being good sports I guess, but after reading this I would never touch a piece of software built by anyone involved with this company. The sheer hubris of their claims combined with the specific reality of their broken garbage software (with the contradicted marketing claims and hallucinated documentation) is literally breathtaking.

semiquaver · Oct 31, 2026 · comment in “Jepsen: ParticleDB 1.1.4”

I don't think that LLMs mirror human cognition or "sanity" as much as people want to believe. We are not constantly hallucinating in any way that applies to the term as commonly understood relative to LLMs or to humans.

krapp · Oct 31, 2026 · comment in “AI isn't replacing SREs, it's deskilling them”

I have been toying with combining offline tools, tiny models and anti-hallucination concepts (RAG, lookup, etc.) I wanted something that would loosely keyword match my documents and help me think through things, whether it was for my next year goals or business ideas, like an autocomplete but matching sources of information and ideas quickly, and allowing me to click for details and quickly go back to my source to u…

shanedevane · Oct 31, 2026 · comment in “Show HN: Offline Tiny-LMs 25MB for personal goals, career and pitch deck advice”

Let’s say you have an incident, and you need a resolution. Do you spend 5 minutes tinking, then solving the problem or send out the AI for the hunt and spend 30 minutes grappling with hallucinations and non-sensical output?

skydhash · Oct 30, 2026 · comment in “AI isn't replacing SREs, it's deskilling them”

> hallucinate an entire operating system every single frame

keeganpoppen · Oct 30, 2026 · comment in “Show HN: A working Nintendo in ~1GB of Qwen weights”

One day a stellar dev and next a bs artist. So this idea that LLM hallucination is not unlike human failings is without merit.

yubblegum · Oct 30, 2026 · comment in “AI isn't replacing SREs, it's deskilling them”

You're comparing apples to oranges there! To hallucinate is arbitrary, you can use a 2B image param model, it will hallucinate an OS no problem* running real time!

flockonus · Oct 30, 2026 · comment in “Show HN: A working Nintendo in ~1GB of Qwen weights”

If the thing’s priced at 2 cents and I can only bid 1 cent or 3 cents, that’s not super useful or informative since it involves a 50% change in value on either side. The new and specific facet of my AI frustration here, is that, true or false or accurate or hallucinated, the analysis leaves me none the wiser about whether I’m totally off base with those priors.

alwa · Oct 30, 2026 · comment in “[dead]”

>spotting the hallucinations

stackskipton · Oct 30, 2026 · comment in “AI isn't replacing SREs, it's deskilling them”

The atrophying skill is one you don't need anymore. The new one - prompting, routing the output, spotting the hallucinations, managing singal/noise etc. - is now where you can add tremendous value.

eqmvii · Oct 30, 2026 · comment in “AI isn't replacing SREs, it's deskilling them”

I like these kinds of projects because it helps to show the massive compute requirement between LLMs and ordinary programs. A while back I saw some people saying we would just have LMMs hallucinate an entire operating system every single frame.

pbohun · Oct 30, 2026 · comment in “Show HN: A working Nintendo in ~1GB of Qwen weights”

yeah i work on this at https://groundedai.company/ we build a product called Veracity focused on citation and claim checking. legal is a major focus for us, also scientific and technical publishing, where we detect hallucinated citations in submitted works and all that jazz

cyclecycle · Oct 29, 2026 · comment in “Zombie stats: Where the numbers AI quotes come from”

The voice is one thing, but I have encountered multiple 'tech support' teams and platforms that will respond to me in pure, unedited LLM output. I know this because the topic is usually an obscure issue, and I have already gone back-and-forth with multiple LLMs to see where their boundary of knowledge is on it, and it will often include hallucinations along the way (e.g. pointing to old/nonexistent settings pathways)

ceroxylon · Oct 29, 2026 · comment in “LLM Voice”

I had been on the pro plan for a few months and recently upgraded to the $200.00 AI Ultra plan because I've gotten good mileage out of 3.8 flash while keeping in mind that it's far from perfect. Argon is better in my experience in terms of reliability (especially in not claiming to have done things it clearly didn't) but I have to agree with the hallucination part, for a "SOTA" model (before Carbon came out at least…

Revanche1367 · Oct 29, 2026 · comment in “Gemini 4 Carbon is the new #1 on the Artificial Analysis Intelligence Index v5”

> has actively hallucinated details

XCSme · Oct 29, 2026 · comment in “Gemini 4 Carbon is the new #1 on the Artificial Analysis Intelligence Index v5”

Makes it seamless for trip planning. Lastly it's the only "SOTA" model in a long time that has actively hallucinated details.

isqueiros · Oct 29, 2026 · comment in “Gemini 4 Carbon is the new #1 on the Artificial Analysis Intelligence Index v5”

> LLMs are always hallucinating that they've found the most severe bug in software history.

dspillett · Oct 29, 2026 · comment in “Zero click, silent, remote Android DoS”

2 things. 1, hallucination is a sensory perception of something.

chomp · Oct 29, 2026 · comment in “New York Times adds more AI features to its Cooking app. Users aren't thrilled”

AI hallucinates.

spl757 · Oct 29, 2026 · comment in “New York Times adds more AI features to its Cooking app. Users aren't thrilled”

Generating an AI hallucination of "called 278 likely voters in the district and it resulted in..." is just as fake as making up the data yourself.

shagie · Oct 28, 2026 · comment in “From confession to conspiracy: How the Wedgewood fake polls story spread”

Nowadays, companies run more tools than they used to, and with time, it will drastically increase. First, because of vibe coding and apps built by agents, which can hallucinate an endpoint or a model name.

stosssik · Oct 28, 2026 · story in “Breaking changes in third-party APIs are turning into a full-time job”

I found it interesting that two of the AI posters in the article have the three-pointed “no kings” crown, and one has a two-pointed version. The AI hallucinated icons from American partisan propaganda into an EU context in a country with a monarchy.

Telemakhos · Oct 28, 2026 · comment in “Barcelona unrest: how to fabricate a protest in the age of TikTok”

Nevermind hallucinations, the donkey-work is important; why bypass the hours of research, which trains our in-built LLM and can lead to new realizations, and get the carrot without much effort?

meristohm · Oct 28, 2026 · comment in “Survival LLM: AI assistant and emergency library that runs on your own computer”

But there still remains the possibility of error/hallucination.

jan_m_savage · Oct 28, 2026 · comment in “Amazon Is Gutting Its AI Division After Sustained Failure”

What about those precious seconds waiting for inference? The danger here is less hallucinations and more, well, everything else about it.

stego-tech · Oct 28, 2026 · comment in “Survival LLM: AI assistant and emergency library that runs on your own computer”

> I have to admit the formalization steps are the fruit of ChatGPT's work and not only would I have not been able to produce them, I still don't understand them and probably never will. So, Hegel is not impenetrable, because ChatGPT can hallucinate something out of his writings, when you tell it to.

Viliam1234 · Oct 27, 2026 · comment in “Nietzschian Debugger”

> the tendency even now of LLMs to hallucinate

IshKebab · Oct 27, 2026 · comment in “Survival LLM: AI assistant and emergency library that runs on your own computer”

The readme on Github does say the system produces hallucinated sources https://github.com/anonnerdfrenzy/survival-llm#how-the-models-did, but really you're correct, for this kind of highly targeted use case the LLM's summary could probably be ignored entirely and the system could work extremely well even just as a search engine: index all of the content up front, chunked by paragraph and maybe enriched with a HyDE s…

macNchz · Oct 27, 2026 · comment in “Survival LLM: AI assistant and emergency library that runs on your own computer”

I don't personally like LLMs for research assistance (yet?) because i can't get over the hallucinating or rather, waste too much time validating that it's not currently overcast when they tell me the sky is in fact blue.

joshAg · Oct 27, 2026 · comment in “The Mythical Man-Month in the Age of AI”

This topic is bringing out the hallucination-panic crowd.

x3haloed · Oct 27, 2026 · comment in “Survival LLM: AI assistant and emergency library that runs on your own computer”

Yeah, without the ability to search the web, a very small model is going to be too dumb to be useful, and maybe even dangerous due to hallucinations.

SwellJoe · Oct 27, 2026 · comment in “Survival LLM: AI assistant and emergency library that runs on your own computer”

That will make it much, MUCH more accurate and truthful than a bare model. But there is still a non-zero chance for it to hallucinate answers, especially when posed a question that it doesn't have any source material in its database.

bityard · Oct 27, 2026 · comment in “Survival LLM: AI assistant and emergency library that runs on your own computer”

If you, however, just drop a wall of text at someone and expect they act on its b/s, this is lazy to the point it borders with microagression. You wouldn't expect a distinguished professor explain basic maths to a student, and we shouldn't expect actual developers address Copilot's stupid, unredacted hallucinations or otherwise unrelated remarks.

cromka · Oct 27, 2026 · comment in “I will no longer upload to NEW”

Would they not likely be more useful and far, far easier and cheaper to create and run (possibly even with far less hardware) than the current "know-all" LLMs? Also might they possibly avoid certain problems ( e.g., hallucinations, deceit, apparently undesirable behavior, etc.) associated with the "know-all" LLMs we now have?

giardini · Oct 27, 2026 · comment in “Sebastian Raschka's from-scratch video course on building reasoning models”

This means total precision and recall. We've found that this approach leads to the lowest rate of hallucination vs. solely targeting agent action verification which has made tremendous strides in capability.

BerendG · Oct 27, 2026 · story in “Show HN: Limma – MCP for total recall and precision in data retrieval for agents”

Harvey care about a potential o(n log n) algorithm, whether or not such an algorithm is contained in OpenAI's paper per se), then it is neither necessary nor (given how it's written) a good idea. As I said in the linked comment, it does not matter whether the idea came from OpenAI's paper, or was hallucinated by Fable, or was arrived at by the reader while reading a potential nonsensical explanation from Fable: the …

svat · Oct 26, 2026 · comment in “The recently proposed o(n log n) integer multiplication algorithm [pdf]”

My browser is underlining "factuality" in red. AI hallucinations or hallucinatiality are still a big problem in my experience.

ape4 · Oct 26, 2026 · comment in “GPT-6.1 Terra”

Not seeing LLM to perform very well here and hallucination would be a major issue (not at all a solved problem, even frontier models do hallucinate/match the most probable path.

raxxorraxor · Oct 26, 2026 · comment in “A 10-year-old CNN outperforms frontier LLMs for skin cancer diagnosis”

I have never used an AI output I didn’t fully understand and that is crucial. It lies and hallucinates and we must police it.

objectively_ran · Oct 26, 2026 · comment in “A mathematician reads the AI future”

As I said in my message, it would determine whether I use the project or not. I need to match it to my use cases, and if I find a match I need to know: is it matching because my use case falls in the domain of what the human collaborator wants from this project, or is my use case a hallucination that will go unsupported?

adastra22 · Oct 24, 2026 · comment in “The Goose Programming Language”

You are definitely overhyping it. LLMs hallucinate.

eek2121 · Oct 24, 2026 · comment in “No signs of AI payoff outside the tech sector”

There were no known breaches or end-to-end exploits as of Friday. I've seen LLM hallucinations purporting to be exploits and real exploits missing pieces (address leaks), but nothing complete.

FreakLegion · Oct 24, 2026 · comment in “Evolving Threat: http2shell is an RCE in Node, Python, and more”

Three years ago AI generated code could basically not be used at all, and I had to rewrite just about every line. Today I have to watch out for generated code making steps in the wrong strategic and architectural direction, and over engineering, no longer hallucinating nonexisting APIs and forgetting to check the auth token.

thomashabets2 · Oct 23, 2026 · comment in “Is this the new cyber normal?”