Google AI Overviews Wrong 30% of the Time on India Flights

A study of 50 India-flight queries found Google AI Overviews gave incorrect baggage rules, wrong visa information, and outdated fares around 30% of the time.

FlightGPT can make mistakes. Confirm flight & fare details before paying.

Google AI Overviews get India flight facts wrong: what to double-check before you travel

By Diya Verma (Diya Verma flies from Tier-2 Indian cities and chases every possible fare hack — reposition flights, hidden-city ticketing, mileage runs and OTA bundle tricks. She has booked 200+ international trips out of Lucknow, Indore and Jaipur.) · Published · 9 min read

HappyFares published a study in 2025 testing 50 India-specific flight queries on Google AI Overviews. Roughly 30% of responses contained a meaningful error — wrong baggage allowance, outdated visa rules, or an incorrect airline policy. As someone who has been burned by acting on overconfident AI summaries before, I found this entirely believable. Here is a breakdown of the error patterns and what you should actually check instead.

TL;DR — the problem in plain terms

Google AI Overviews — the AI-generated summary boxes that appear above search results — have been giving incorrect answers on India-specific flight queries at a meaningful error rate. A 2025 study by HappyFares tested 50 common queries and found around 30% returned incorrect or misleading information. The errors concentrated in three areas: IndiGo baggage allowances (where policy changes frequently), visa and documentation rules (which vary by nationality and update regularly), and airline-specific policies that AI models have not been retrained on. The fix is not to stop using AI — it is to know which facts to verify on the official source, especially before spending money or packing your bags based on what an AI told you.

What HappyFares actually found — the categories of errors

The study was not peer-reviewed academic research, but it is the most systematic India-specific test of AI Overview accuracy I have seen published. The broad findings, paraphrased from their reporting:

What the study also found: when the query was very specific and recently published information existed (e.g., 'Air India direct Delhi to San Francisco route'), the AI Overviews were more accurate. The errors concentrated on detail-level, policy-specific questions where the 'correct' answer changes regularly.

Why IndiGo baggage policy is the biggest AI pitfall for Indian travellers

IndiGo's baggage structure is genuinely confusing, and it changes more often than travellers realise. Here is the current framework — verify on IndiGo.com for the latest, because this is precisely the kind of detail that AI gets wrong:

When an AI Overview says 'IndiGo allows 15 kg free baggage,' it is wrong for most current base fares. When it says 'IndiGo charges ₹X for extra baggage,' that figure may be out of date within months. I have seen travellers show up at IndiGo check-in counters with a printout of what Google told them the baggage fee was, only to pay a substantially higher airport fee. The AI was not lying — it was trained on older data.

The correct move: always check IndiGo's official baggage page (IndiGo.com/baggage or equivalent) for the current allowance and add-on fee structure before flying. Same principle applies to Air India, Air India Express, and Akasa.

Visa rules: the category where being wrong is highest-stakes

Visa errors in AI Overviews are the most dangerous category for Indian travellers — because acting on wrong visa information can mean being denied boarding, turned back at the border, or worse.

Common error patterns I have personally observed (not from the HappyFares study, from my own searches over the past year):

The general principle: for any visa question that affects whether you can board a plane, never act on an AI Overview or any AI tool as your sole source. Cross-check on the official embassy website, the VFS/TLScontact page, and ideally the IATA Travel Centre (which airlines use). For a starting point on visa research, FlightGPT's visa panel links to current official sources for common India-departing routes.

Why AI Overviews get India-flight facts wrong more than they should

It is worth understanding why this happens, because it affects how you should weight AI-generated information on time-sensitive topics:

What to actually check — a quick reference list

Here is my personal checklist of things I always verify on the official source before flying, regardless of what any AI tool told me:

For fare searching and comparison, AI tools like FlightGPT are built for fare accuracy — the search results reflect live inventory. But policy information (baggage, visa, entry requirements) is a different beast, and no AI tool should be your sole source for anything that determines whether you get on the plane.

Bottom line

AI Overviews are useful for general orientation — 'what are the top airlines on this route' or 'what is Goa's weather in February' — but they are a liability for time-sensitive policy questions. The 30% error rate on India-flight queries from HappyFares' study is not a reason to panic about AI; it is a calibration nudge. Know which questions need an authoritative source and go there directly. See also our related pieces on testing Google Flights' own AI deals feature and how the Skyscanner + ChatGPT integration compares.

Frequently asked questions

Are Google AI Overviews reliable for India flight information?

For general, stable information (major airlines on a route, approximate flight duration) they are often accurate. For volatile policy details — IndiGo's current baggage fee structure, visa requirements for Indian passport holders, specific refund policies by fare class — they have a meaningful error rate. A 2025 HappyFares study found around 30% of India-specific flight queries returned incorrect or misleading information. Always verify policy details on the airline's official website or the relevant embassy page.

What is IndiGo's current baggage allowance for domestic flights?

IndiGo's domestic base fares typically include 7 kg cabin baggage and zero free check-in baggage — check-in bags must be added for a fee. The fee varies by weight option (15 kg, 20 kg, 25 kg) and is cheapest when added at booking. Airport counter rates are the most expensive. These details are subject to change — always verify on IndiGo.com before your flight, as this is exactly the kind of policy that AI Overviews have gotten wrong.

Is Thailand visa-free for Indian passport holders in 2026?

Thailand extended a visa exemption for Indian passport holders in 2023–2024 that allows a stay of up to 60 days without a visa or visa-on-arrival fee. However, entry conditions (port of entry, proof of funds requirements, return ticket requirements) still apply and can change. Verify the current policy on the official Royal Thai Embassy website or the Tourism Authority of Thailand before booking, as this is a policy area where AI Overviews have given conflicting and outdated information.

Which source should I trust for airline baggage rules?

The airline's own official website (IndiGo.com, airindia.com, akasaair.com, spicejet.com) is the only authoritative source. OTA summary pages, AI Overviews, and travel blog articles all lag behind policy changes. The IATA Travel Centre (iatatravelcentre.com) is the authoritative source for visa and documentation requirements — it is the same database airlines use to determine whether to allow you to board.

Why does Google AI Overview cite the wrong IndiGo baggage policy?

AI language models are trained on web data up to a cutoff date, and IndiGo's baggage policy has changed multiple times in recent years. The model may reflect an older policy version. Additionally, a large volume of OTA and travel blog content repeats older or inaccurate policy summaries, and AI models can retrieve and synthesise those instead of the airline's current official page. This is a training data quality and retrieval problem — not a fundamental flaw in AI, but a real limitation on volatile policy topics.

Should I use AI tools at all for India flight planning?

Yes — for fare search, date flexibility, and route comparison, AI flight search tools are genuinely useful and operate on live inventory data. FlightGPT is built for that purpose. The distinction is between fare search (where live data tools are accurate) and policy/documentation questions (where you must go to official sources). Use AI for 'how much does this flight cost and when is it cheapest' — and official sources for 'what documents do I need and what baggage is allowed.'