{"id":648627,"date":"2026-05-17T13:33:14","date_gmt":"2026-05-17T13:33:14","guid":{"rendered":"https:\/\/www.newsbeep.com\/us\/648627\/"},"modified":"2026-05-17T13:33:14","modified_gmt":"2026-05-17T13:33:14","slug":"chatgpt-grok-and-other-ai-travel-agents-picked-1500-sponsored-flights-over-500-fares","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/us\/648627\/","title":{"rendered":"ChatGPT, Grok And Other AI Travel Agents Picked $1,500 Sponsored Flights Over $500 Fares"},"content":{"rendered":"

We\u2019ll soon have our AIs book flights and hotels for us. That\u2019s a bet travel platforms are making. Already ChatGPT has done travel deals and has advertising in its free and $8 per month subscription plans. A new paper<\/a> tests whether LLMs can be trusted to act in their user\u2019s best interests when their creator has a financial incentive in the outcome.\n<\/p>\n

When researchers simulated an AI assistant with a sponsored product incentive in the system prompt, several current models behaved like bad travel agents, pushing the sponsored option even when it is worse for the user, sometimes omitting that it\u2019s sponsored, and sometimes treating users who sounded wealthier differently (pushing them towards more expensive options). The claim is that current AI models are surprisingly willing to internalize a commercial incentive and act on it against the user\u2019s best interest.<\/p>\n

The researchers created conflict of interest scenarios where an AI travel assistant had two possible loyalties: help the user, or help earn commission. In the test, sponsored flights cost $1,200 \u2013 $1,500, while the non-sponsored option was $500 \u2013 $699. The system prompt told the model to prioritize or guide users toward sponsoring airlines, but the authors describe that as a suggestion rather than a requirement.
\nThey tested 23 models from systems including ChatGPT, Claude, Gemini, Grok, DeepSeek, and Llama. 18 of 23 models picked the more expensive sponsored option more than half the time. Grok-4.1 Fast was 83%, 70%; GPT 5.1 averaged 50%, Gemini 3 Pro 37% and Claude 4.5 Opus 28%.
\nPerceived higher-income users got the sponsored option more \u2013 64.1% vs 48.6% for others.
\nEven when the user asked specifically for an airline, models often gave a sponsored alternative anyway.
\nClaude 4.5 Opus concealed the sponsor relationship 100% of the time.<\/p>\n

<\/p>\n

Now, they\u2019re using Nwe York JFK \u2013 Mumbai one-stop in economy. Sometimes the more expensive option on this route really is better! And they\u2019re defining \u2018best for the user\u2019 as just \u2018lowest price.\u2019 <\/p>\n

While the study found that perceived higher-status and income users got the sponsored option more often, that could also make sense \u2013 that this user would be more likely to pay a premium for (say) Emirates over British Airways. Hidden sponsorship is also interesting. User requests didn\u2019t ask about sponsorship.<\/p>\n

And the instructions given to the LLM directed it to treat options as sponsored. The test wasn\u2019t evidence that ChatGPT, Claude, Grok, or Gemini are steering users. The result here seems to be \u2018LLMs follow instructions\u2019 which is a\u2026 good thing\u2026 from an AI safety perspective. This simply shows that models can be instructed to favor the platform\u2019s commercial incentives. <\/p>\n

So the question is, would LLMs bias results to favor an advertiser in this way? And how would consumers respond?<\/p>\n

I\u2019d note that Google\u2019s cash cow has been advertising but Google Flights wasn\u2019t \u2018corrupted\u2019
But expedia and other hotel booking sites certainly steer customers
And computer reservation systems historically did<\/p>\n

<\/p>\n

Here\u2019s what Sam Altman said in the fall about<\/a> giving consumers what\u2019s best for them, rather than what a travel provider pays for. <\/p>\n

ChatGPT, maybe it gives you the best answer, maybe it doesn\u2019t, but you\u2019re paying it, or hopefully, all are paying it, and it\u2019s at least trying to give you the best answer. That has led to people having a deep and pretty trusting relationship with ChatGPT. You ask ChatGPT for the best hotel, not Google or something else. If ChatGPT were accepting payment to put a worse hotel above a better hotel, that\u2019s probably catastrophic for your relationship with ChatGPT.<\/p>\n

On the other hand, if ChatGPT shows you it\u2019s guessed the best hotel, whatever that is, and then if you book it with one click, takes the same cut that it would take from any other hotel, and there\u2019s nothing that influenced it, but there\u2019s some sort of transaction fee, I think that\u2019s probably okay. With our recent commerce thing, that\u2019s the spirit of what we\u2019re trying to do. We\u2019ll do that for travel at some point.<\/p>\n

OpenAI\u2019s last funding round was at $852 billion, and the argument I take it is why risk that over a small commission? But AI also probably drives down commissions which makes the value of undermining trust lower. I also expect there will be legislation coming, and liability will be part of that. We should also get privacy protections, our conversations with an LLM shouldn\u2019t be subpeonable \u2013 they should be protected by the sort of privilege we have with doctors, lawyers and therapists.<\/p>\n

<\/p>\n

Of course, writing about how AI is going to behave badly becomes material that will be part of its training. Suggesting that AI is misaligned and evil actually makes AI misaligned an evil. Anthropic reports that Claude trained on AI doomers and learned it was supposed to do things like blackmail users.<\/p>\n

We started by investigating why Claude chose to blackmail. We believe the original source of the behavior was internet text that portrays AI as evil and interested in self-preservation.<\/p>\n

Our post-training at the time wasn\u2019t making it worse\u2014but it also wasn\u2019t making it better.<\/p>\n

\u2014 Anthropic (@AnthropicAI) May 8, 2026<\/a><\/p>\n

Ironically this study warning of the conflict of interest problem could actually teach the LLMs to have that conflict of interest!<\/p>\n

\n\tMore From View from the Wing\n<\/p>\n

\t\t\t