Google’s AI Mode Adds Global Flight Tracking and US Hotel Booking

01Google Takes AI Travel Planning Into Transactions: Global Flight Price Tracking and Direct Hotel Booking for US Users

Google is extending AI Mode in Search from recommending trips to helping travelers act on those plans. AI Mode already used real-time flight and hotel data, web recommendations and itinerary generation; it can now track airfare, show loyalty-points prices and, for some US users, complete hotel reservations.

The three additions move different parts of travel planning closer to a transaction, but their availability and checkout paths differ. Flight tickets and rewards bookings still require users to leave Google, while eligible hotel reservations can be paid for within the AI Mode flow.

For flights, users can describe a route and travel dates in conversation, review current options, and ask AI Mode to monitor the prices. The results draw on data from more than 300 partner airlines and travel sites. After a user confirms an alert, Google sends an email when prices change for the specified destination and dates.

That price-tracking feature is available in more than 180 countries and territories. AI Mode can also help construct a full flight itinerary, but it does not sell the ticket directly: travelers must complete the purchase on an airline website or their preferred booking platform.

A second feature addresses the difficulty of comparing cash prices with loyalty rewards. AI Mode can now display how many points or miles are required for matching flights and hotels, using current data from participating rewards programs. This lookup is available globally, but redemption still happens through a link to the partner’s site.

Initial rewards partners include Alaska Airlines/Hawaiian Airlines, American Airlines, Choice Hotels International, Hilton, and Wyndham Hotels & Resorts. Google says support for Accor, Flying Blue, Hyatt, LATAM Airlines, and Lufthansa Group will follow in the coming weeks.

The most complete transaction flow is hotel booking, which has started rolling out in English in the United States. Users can compare visual hotel listings, guest reviews and other factors, then choose “Continue on Google” for an integrated partner. They can select a room type, review details including the cancellation policy, and pay through Google Pay.

The rollout covers Booking.com, Choice Hotels International, Expedia, Hilton, Hotels.com, IHG Hotels & Resorts, Marriott International, Priceline, Trip.com, and Wyndham Hotels & Resorts. The hotel or booking platform remains the merchant of record and handles customer service. Google has not said when this booking capability will expand beyond the United States.

02Nvidia Emerges as Potential Hugging Face Buyer in Talks at a Valuation Above $13 Billion, but No Agreement Has Been Reached

Nvidia has held acquisition talks with Hugging Face in recent weeks over a potential deal valuing the company at more than $13 billion, according to a person familiar with the matter. Hugging Face is a central platform for open-source artificial intelligence, hosting millions of models and datasets that developers use to build applications.

The discussions would mark a significant change in Nvidia’s relationship with the company: from investor to potential owner. The parties have not reached an agreement, the person said, and the talks could still collapse. Neither Nvidia nor Hugging Face responded to requests for comment, leaving the proposed price, transaction structure and any regulatory arrangements unknown.

Nvidia participated in Hugging Face’s $235 million funding round in 2023, when the company was valued at $4.5 billion. Late last year, Hugging Face rejected a separate Nvidia offer to invest $500 million at a $7 billion valuation. Hugging Face said at the time that it did not want a dominant investor capable of influencing its decisions.

A takeover at more than $13 billion would therefore value Hugging Face at nearly three times its 2023 level and give Nvidia control rather than a minority investment. Nvidia has the resources for large transactions: the chipmaker said it had committed $18 billion to equity investments for the remainder of its fiscal year, in addition to $47.9 billion already held in private companies.

Microsoft also met with Hugging Face, but two people familiar with the matter said those talks are no longer continuing.

Control of Hugging Face could deepen Nvidia’s access to developers working with open-source models and potentially bring more computing workloads to its chips. It would also put the platform’s hardware neutrality under scrutiny. Hugging Face currently supports technology from across the industry, including Nvidia competitors AMD and Intel, and that broad compatibility is one of its strengths.

There is no indication yet whether Nvidia would change that support if a deal proceeds. The companies have not signed an agreement, and the final price, structure and treatment of competing hardware remain unresolved.

03Study Finds 1,655 “Ghost Author” Records on Zenodo, With Fictional Papers Entering Academic Indexes Through Real DOIs

Researchers from Samsung and the University of Warsaw say recurring fictional names generated by large language models have become clues for tracing AI-made material as it spreads into the scholarly record. Their findings appear in a preprint, meaning the study has not been presented as peer-reviewed journal research.

The researchers call these nonexistent people “ghost authors.” Names including Elena Vasquez and Marcus Chen have appeared as experts, authors or characters across hundreds of independently produced AI-generated documents. The repetition occurs because some models default to particular names in certain contexts and sometimes generate correlated groups of names that tend to appear together.

The team used those patterns to search academic-paper databases for names already known to recur in model output. On Zenodo, a repository operated by CERN, it identified 1,655 records attributed to ghost authors. According to the preprint, the records claimed to come from journals that did not exist, included fabricated publication dates and were often backdated relative to their upload dates.

Zenodo allows anyone with a free account to create a Digital Object Identifier, or DOI, for an uploaded record. A DOI is a persistent string used to identify scholarly material, but receiving one does not mean a paper passed peer review or was formally published by a journal. The researchers said the suspect records nevertheless carried genuine DataCite DOIs that scholarly aggregation services could collect.

That infrastructure can give fictional scholarship a wider footprint. The researchers found ghost names on ResearchGate in synthetic research groups whose supposed collaborators appeared to draw on several model families. They also said Google Scholar and Semantic Scholar indexed such material without verification, although the source does not establish how many of the 1,655 Zenodo records reached each platform or independently verify how every record was generated.

The pattern extends beyond repositories. Research Gold, a company previously investigated by 404 Media for offering purportedly human medical research that was entirely AI-generated, listed Elena Vasquez as its founder and lead methodologist. It removed the name after the article appeared.

The researchers believe name frequencies may help identify which model family produced a document; Elena Vasquez, for example, was especially common in Claude Sonnet 4 output. But that signal may decay quickly as AI-generated pages enter the web and feed back into models’ training data, leaving the method’s long-term accuracy unknown.