From safety-first to sales pitches, AI companies speak for themselves

This article is part of a series, Bots and ballots: How artificial intelligence is reshaping elections worldwide, presented by Luminate.
Wherever you look, tech giants want to be seen doing the right thing when it comes to artificial intelligence.
Meta and OpenAI published lengthy treaties detailing how they’re protecting this year’s global election cycle from harm. New kids on the bloc — France’s Mistral and the United States’ Inflection AI — espoused their “safety first” approach to the emerging technology. Old-school titan Amazon wanted everyone to know it was building its own AI systems responsibly.
Amid such policy jargon, it’s easy to get lost.
POLITICO set out to cut through the noise to determine the leading AI companies’ talking points. Many of these lines were front and center when firms’ executives set out to convince lawmakers worldwide that they could be trusted with fast-paced AI innovation.
To that end, we pulled together numerous AI-related public policy documents, technical specifications and company terms and conditions from nine of the West’s leading artificial intelligence companies over the last two years. These included statements from OpenAI, Meta, Amazon, Alphabet and Microsoft, as well as their smaller rivals Mistral, Aleph Alpha, Inflection AI and Anthropic.
We then crunched the data to measure the raw frequency of words commonly appearing in AI-related statements from these companies over that time period. The goal: figuring out, when push came to shove, these companies’ exact repeated talking points, and how they differed.
Not every company had made the same number of public statements. So we limited the data gathering to a maximum of the 100 most common words, and then edited the results to include two-word combinations (like generative AI) and removed stock language (like companies’ addresses) that could skew the results.
What appears below is a clear picture of how each company approaches the ongoing global lobbying battle around artificial intelligence, based solely on their public statements from 2022 onward.
More than anything, the data shows a repetition of clear policy jargon — “responsible AI,” “frontier models,” “AI governance” — that lies at the heart of each company’s lobbying. Most of that is indecipherable to those not working on the front line of how countries, companies, academics and civil society groups are trying to corral the technology into some form of structure.
Interestingly, in this year of global elections, terms associated with voting protection and integrity did not play a central role in companies’ public statements.
The firms are listed in alphabetical order.
This German AI company may not be a household name compared with some of its larger American rivals. But the Heidelberg-based tech firm has pitched itself as Europe’s homegrown champion. It’s played up its credentials in so-called sovereign AI, an effort to offer the Continent an alternative to relying on foreign providers of the emerging technology.
As one of the world’s most important tech titans, the parent company of Google, YouTube and Gemini — its in-house rival to OpenAI’s ChatGPT — has tried to talk to as many parts of the AI world as possible. That includes promoting AI safety to policymakers, inclusivity to civil society groups, and innovation and research to potential customers for its latest advances.
The Seattle-based e-commerce giant is renowned for its penny-pinching ways and obsessive focus on its customers above anything else. So it’s not surprising that Amazon, which doesn’t exactly first come to mind for many for AI innovation, has primarily focused its public statements on promoting its technology to speed up cloud computing and its other geeky products.
For an AI startup that’s raised about $4.5 billion since 2021, Anthropic is hardly an also-ran in this digital Great Game. But, based on its public statements, the company started by former OpenAI executives has pitched itself as the responsible voice in AI governance. That has given it a seat at the policymaking table alongside larger names like Meta and Microsoft.
There’s an open question about what will happen to this AI startup, now that one of its co-founders, Mustafa Suleyman, has jumped ship to run Microsoft’s rival — and bigger — business. But over two years, the California-based company has heavily promoted its AI-powered personal assistant to differentiate itself in an already crowded market.
The parent company of Facebook, Instagram and WhatsApp is no stranger to bruising global policy fights. But Mark Zuckerberg has positioned his company slightly differently from rivals. That includes centering Meta’s AI development on so-called open-source technology, systems that can be readily harnessed by the many, not just a select few.
The maker of Word and Outlook isn’t messing around with its message on AI. In scores of documents, many of which were aimed directly at policymakers and politicians, Microsoft overdid it on key buzzwords like “responsible AI,” “risk management,” and “AI governance.” Of all the companies surveyed, the American tech giant got the most mileage out of such jargon.
When you think of France’s approach to technology, top-down regulation, not grassroots innovation, may first come to mind. But Mistral — fresh off deals with both Microsoft and Amazon — has been keen to promote a business-first stance when publicly talking about its work.
As the company that ignited the most recent AI craze with the release of its ChatGPT product in late 2022, OpenAI has understandably been in the crosshairs of officials and tech rivals alike. That’s led to a surge of geeky research publications detailing how its systems were developed, as well as a greater focus than many rivals on reducing the potential downsides to AI.
This article is part of a series, Bots and ballots: How artificial intelligence is reshaping elections worldwide, presented by Luminate. The article is produced with full editorial independence by POLITICO reporters and editors. Learn more about editorial content presented by outside advertisers.