Vass Bednar: Meta’s latest audacious move shows why Canadians must keep pushing back against Big Tech

A couple of months ago, when speaking at the European Parliament amid talk of partnership and shared prosperity, Prime Minister Mark Carney delivered a warning about big tech.

“Technology platforms increasingly seek to act as sovereign powers,” Carney said. “They want to dictate the rules of regulation — and even circumvent them. They want to control our data, spread misinformation and steal our children’s childhoods.”

As if it was trying to prove his point, earlier this month, Meta Platforms Inc. informed business customers that it was updating its commercial terms of service with language that many believed gave Meta clearance to crawl their websites and use the content to train its AI products.

After a swift backlash, particularly from publishers in the U.S., Meta removed the word “train” from the new terms within a day, claiming that the crawling applied only to its ad and business products.

But the backlash hasn’t fully died down. Some experts say the updated terms of service — which allow Meta to “develop” and “improve” its business products using crawled data — could still give it unfettered access to publishers’ content for AI training.

Meta has now delayed the implementation of the new terms until Nov. 20, which the company has said will give it more time to explain how it uses web crawlers and delineate “what’s changing and what’s not.”

It’s fitting to think that this episode is happening as Aaron Sorkin’s The Social Reckoning plays to disinterested theatre-goers. The film dramatizes the 2021 Facebook Files, a series of articles published in the Wall Street Journal based on leaks from Facebook whistleblower Frances Haugen, which galvanized readers as it was revealed that the company knew its products caused harm. On screen and off, the pattern appears to be continuing: Meta moves first and fast, and then “asks” permission when they are caught.

Meta’s crawlers already roam the open web gathering training data.

The company also trains on public posts, photos and comments on platforms it owns, Facebook, Instagram and Threads (going back to 2007). Privacy laws force a formal opt-out option in the EU and the U.K., but elsewhere, the main defence for individuals is making your posts private.

Under the original wording of its new terms of service, critics feared that any business using Meta’s ad or business tools would have agreed, by contract, to allow Meta to use all of their content to train its various products, and that the new terms could override or circumvent previous objections.

There is a certain audacity to flipping every one of your existing advertising relationships into a free license to feed your AI, and in treating anyone who shares anything on your services as a source of free training material.

Canadian news publishers are already familiar with Meta’s strongarm tactics, as the company has blocked news for the past three years, rather than pay publishers for sharing their content on its platforms.

When it comes to AI, Meta is in the same frantic race as other frontier developers. The firm expects to spend $135 billion to $145 billion on AI infrastructure this year, and that’s just for data centres. If Meta had to pay for training data, the economics of the technology might never work.

In May, federal and provincial privacy commissioners found that OpenAI violated Canadian privacy law, partly by scraping public websites and social media without valid consent. That finding should raise questions about Meta’s own crawling.

Perhaps a deeper issue is why we continue to use Meta’s platforms — voluntarily creating content for the machine to ingest and monetize. We continue to create the raw material that informs algorithmic models for a company that refuses to pay for the journalism it takes, or even show it to us.

The federal government spent $1,409,238 on Facebook and Instagram ads in fiscal 2024-25. Though well below historic highs, that’s nearly triple the $476,271 spent in 2023-24. The platform blocks traditional news media, and yet we’re still providing it with tax dollars.

One can expect Meta to continue to creatively rewrite its terms as it attempts to secure reliable profitability. Why do we keep agreeing? Governments and courts are working to set terms of their own, and Meta’s latest walk back is a reminder that pushback works. But this fight is far from over.

Muse, Meta’s surveillant Tamagotchi-like AI agent demands deep trust. But you don’t give your car over to a valet that you just caught rifling through the glove box. The latest episode is a reminder that we can say ‘No’ to Meta. We should keep saying it.

Vass Bednar is managing director of the Canadian Shield Institute