Fire Sale Firestarters
One surprise beneficiary of the Spirit Airlines bankruptcy debacle: Google.
Google spent $10 million to acquire “a trove of deidentified business data, software code and operations records” from the now-defunct airline, Bloomberg Law reports.
The data stockpile Google bought for a song includes 100 million emails, 500 million Microsoft Teams chats and collaboration records, plus information related to revenue, aircraft operations, employee productivity audits and fraud. Notably, however, the deal doesn’t include addressable passenger info or loyalty program profiles.
The second-highest bidder was Mercor.io, an AI talent recruitment company that offered $7.5 million.
The Spirit Airlines fire sale is yet another example of the insatiable appetite that LLMs have for data.
404 Media just reported the fascinating tale of how it used an AirTag to track a rare book to an Amazon facility where books are despined, scanned and then discarded. The Information, meanwhile, reported on OpenAI, Anthropic and other LLM operators buying up enterprise software logs, internal messages and Slack communiques from failed startups, sometimes for sizable sums considering the businesses are toast already.
But the data “can be helpful in improving our products and AI models,” as Google says of its Spirit Airlines data acquisition.
Ad Hallucinations
This is a big week for OpenAI. It expanded advertising to Mexico and Brazil, and is planning to roll out ads across Europe by the end of the month.
There’s also a new ChatGPT mode for teenagers with parental consent and built-in safeguards. And it announced an overhaul of internal safety protocols, because its latest LLM model, called Astra, demonstrated a worrisome tendency to breach its testing sandbox containment and make unauthorized changes.
But behind all these headlines, something a little strange is happening. ChatGPT ad units have been documented – and anecdotally observed all over the place – doing a weird switcheroo.
As Search Engine Roundtable put it, “you might notice” that ChatGPT ads have been loading momentarily with one advertiser before swapping out a new one, which remains set. The odd bug brings up all sorts of uncomfortable questions. Is ChatGPT double-dipping on those ad units by charging two brands? Are the brands that serve the last ad aware that some other brand briefly flitted into the ad unit before them? Is the ghost advertiser even aware their ad was served at all?
The online ad industry already has a measly standard for what constitutes an ad impression: just half of the pixels in view for one continuous second.
ChatGPT’s response, apparently, is: “Hold my beer.”
The World’s Worst Mirrors
If a platform is flooded with right-wing content, it’s intuitive that the users who engage with that platform would likely have similar values.
But on X, that isn’t the case.
A recent study published in the “Proceedings of the National Academy of Sciences” found that while X users generally follow accounts whose content reflects their stated values, the platform’s algorithm often amplifies posts that don’t – especially for Democrats.
“It doesn’t matter who you follow or what your stated values are,” 404 Media reports. “X reads replying as engagement and will send more of the infuriating posts the user’s way.”
That makes engagement a particularly slippery signal, not only of what people agree with but of what they may be angrily reacting to.
Meanwhile, X is trying to win back advertisers after many left then-Twitter following Elon Musk’s takeover of the platform. For the advertisers that remain, the challenge is figuring out what their target audience actually wants to see because, in this case, engagement is not a reliable measure of interest.
And X isn’t the only platform making it harder to distinguish real attention from low-quality activity. YouTube caused a stir this week by changing how it counts video views, now rewarding cheap, unengaged views rather than gauging whether viewers actually stick around to watch.
But Wait! There’s More!
PepsiCo is conducting an AI marketing transformation agency review. [Ad Age]
The IAB releases the second version of its AI Transparency and Disclosure Framework. [release]
Disney-owned ABC filed a First Amendment lawsuit against the FCC, claiming that the agency starting an early renewal process for ABC’s broadcast license is retaliation for programming critical of President Donald Trump. [CNBC]
YouTube is in advanced talks with the NBA to secure rights for a streaming service that would aggregate the local broadcasts of almost all of the teams in the league. [Sports Business Journal]
You’re Hired!
Digital health company PatientPoint hires Jeremy Hlavacek as GM of advertising. [post]
Thanks for reading AdExchanger’s Daily News Roundup. Want it by email? Sign up here.
