As a newsroom executive, I have long tried to resist appeals to vet donors to nonprofit journalism for their politics or even the savoriness of their business practices. When you do that, you raise questions about whether you could fairly cover any potential donors on your banned list, and you also run the risk of offering some sort of implicit endorsement to the rest. Both of those perceptions are antithetical to the independent stance that good journalism requires.

To take one notorious example, we never received a check from the Koch family’s interests when I was running the business side of ProPublica, but I would have been happy to take it if we had. Nor am I one of those people who think that it makes sense to refuse corporate contributions. Almost all major donors derive the money they are offering from a successful corporation.

With that said, I want to offer some cautionary thoughts on the largesse from the company that has become the leading corporate supporter of American journalism, OpenAI. In gifts through its People-First AI Fund, and in large grants disbursed through the American Journalism Project and the Lenfest Institute for Journalism (both occasional consulting clients of mine), OpenAI has distributed $25 million in cash to U.S. newsrooms in the past few years, a large multiple of the journalism grants from Microsoft and Google.

OpenAI has also made a number of significant one-off gifts, so its total is considerably larger. Anthropic has not yet joined this party. (The Atlantic has a content-licensing agreement with OpenAI, although it does not receive grant funding from the company.) I know that the money is needed, and I am not questioning the uses to which it is currently being put. I think newsrooms need to be more creative in the use of AI, not less; more innovative; and less afraid of the implications for particular kinds of work that may become outmoded. Here’s what I think the recipients need to bear in mind.

The company’s motivation here is not altruistic. OpenAI says that its intent is “to benefit all of humanity,” but we know that some people there and elsewhere worry that it is, in fact, going to end up snuffing out humanity. We also see that, in the face of such warnings, the company races along, its CEO announcing that “the world should accept some bad things happening.”More narrowly, the firm’s spending on supporting news is almost certainly seen internally as a form of marketing and public relations, helping to spur product adoption and perhaps blunt any impulses toward regulation or taxation, both of which seem to be crucially and urgently necessary.

The money may be material to us, but it isn’t to them. It should be sobering, for instance, that OpenAI spent roughly the same amount of money trying to intervene in a single Manhattan congressional primary this past spring as it has in newsroom funding over the years through Lenfest.

Our worldviews are at fundamental variance. The core principles of journalism focus on the search for accuracy with fairness and the discovery of truth. The essence of the business of generative artificial intelligence seems to be the maximization of profit, whether or not other goals are achieved.

The spigot could shut off at any time, and likely eventually will. We have been down this road before with tech firms, and need to assume that this funding won’t be sustained, and shouldn’t be part of any plan for sustainability. Facebook was a significant funder of news until it not only wasn’t, but then took steps to choke off the flow of news to its users. Google purported to be especially solicitous of news until, in my view, it detected insufficient gratitude and greater support for regulation (especially in Canada) and largely reversed course. (The Atlantic has received grant funding in the past from Facebook and Google but no longer accepts such grants to fund its journalism.) Inside the velvet glove is a mailed fist.

Journalists should never forget that there is significant evidence that OpenAI has built its enormous machine in small but significant (at least to us) part by appropriating our intellectual property without compensation. That’s a nice way of saying that the company is arguing in court that it has the right to take our lifeblood while laying waste to important parts of our business.

This is ultimately a legal question, but one in which the company—and, to be sure, its competitors at Anthropic, Google, Meta, and Microsoft—is our adversary. The stakes in those fights may be existential for newsrooms; they will not be for the platforms.

Transparency is always a good idea. When creating new AI tools or devising new methods or products using AI that are ultimately funded by an AI company, journalists owe it to their readers, listeners, and viewers to let them know about that funding. Don’t limit the disclosure to an intermediary source through which the money passed.

Innovation needs to be coupled with accountability. Even as OpenAI and, to a lesser extent, some of its competitors are paying newsrooms for more rapid and clever adoption of their products and services, it’s very important that anyone taking this funding also invest commensurately in increased accountability for these behemoths and the havoc they are wreaking in some local communities, many funding streams, and, moving forward, job markets. I do have faith in the ability of editors and publishers to maintain editorial independence from funders; I have seen it in action many times.

But the larger the portions of food on offer, the greater the temptation not to bite the hand offering it. We need to help one another to resist this, including by pointing out compromises among colleagues if and when we see them.

One important caveat: I am far less concerned about the explosion almost certainly ahead in individual philanthropy derived from AI than I am about the direct provision of corporate funds. There may be issues with some individuals, of course. But the multiplicity of likely funders will help, as will their varying motivations.

The truth is that corporate philanthropy has always been something of an oxymoron. The duty of corporations is to their shareholders, and specifically to the maximizing of shareholder value. Journalists can take some comfort that, in companies with more enlightened management, this objective is sought over the long run, which may more closely align with societal interests. Managers with more limited vision, as well as those so enamored of their own capabilities as to conflate society’s objectives with their own, present a less happy case. All of us in the news business would do well to look hard at which of these types we believe are leading the AI revolution today.