This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.
The AI sector is plagued by a lack of competition and a lot of deceit—or at least that’s one way to interpret the latest flurry of actions taken in Washington.
Last Thursday, Senators Elizabeth Warren and Eric Schmitt introduced a bill aimed at stirring up more competition for Pentagon contracts awarded in AI and cloud computing. Amazon, Microsoft, Google, and Oracle currently dominate those contracts. “The way that the big get bigger in AI is by sucking up everyone else’s data and using it to train and expand their own systems,” Warren told the Washington Post.
The new bill would “require a competitive award process” for contracts, which would ban the use of “no-bid” awards by the Pentagon to companies for cloud services or AI foundation models. (The lawmakers’ move came a day after OpenAI announced that its technology would be deployed on the battlefield for the first time in a partnership with Anduril, completing a year-long reversal of its policy against working with the military.)
While Big Tech is hit with antitrust investigations—including the ongoing lawsuit against Google about its dominance in search, as well as a new investigation opened into Microsoft—regulators are also accusing AI companies of, well, just straight-up lying.
On Tuesday, the Federal Trade Commission took action against the smart-camera company IntelliVision, saying that the company makes false claims about its facial recognition technology. IntelliVision has promoted its AI models, which are used in both home and commercial security camera systems, as operating without gender or racial bias and being trained on millions of images, two claims the FTC says are false. (The company couldn’t support the bias claim and the system was trained on only 100,000 images, the FTC says.)
A week earlier, the FTC made similar claims of deceit against the security giant Evolv, which sells AI-powered security scanning products to stadiums, K-12 schools, and hospitals. Evolv advertises its systems as offering better protection than simple metal detectors, saying they use AI to accurately screen for guns, knives, and other threats while ignoring harmless items. The FTC alleges that Evolv has inflated its accuracy claims, and that its systems failed in consequential cases, such as a 2022 incident when they failed to detect a seven-inch knife that was ultimately used to stab a student.
Those add to the complaints the FTC made back in September against a number of AI companies, including one that sold a tool to generate fake product reviews and one selling “AI lawyer” services.
The actions are somewhat tame. IntelliVision and Evolv have not actually been served fines. The FTC has simply prohibited the companies from making claims that they can’t back up with evidence, and in the case of Evolv, it requires the company to allow certain customers to get out of contracts if they wish to.
However, they do represent an effort to hold the AI industry’s hype to account in the final months before the FTC’s chair, Lina Khan, is replaced when Donald Trump takes office. Amid all the nominations in recent weeks, the FTC looks to have a far smoother transition of leadership ahead than most other federal agencies. On Thursday, Trump announced that he’d picked Gail Slate, a tech policy advisor and a former aide to vice president–elect JD Vance, to lead the agency. Trump has signaled that the FTC under Slater will keep tech behemoths like Google, Amazon, and Microsoft in the crosshairs.
“Big Tech has run wild for years, stifling competition in our most innovative sector and, as we all know, using its market power to crack down on the rights of so many Americans, as well as those of Little Tech!” Trump said in his announcement of the pick. “I was proud to fight these abuses in my First Term, and our Department of Justice’s antitrust team will continue that work under Gail’s leadership.”
That said, at least some of Trump’s frustrations with Big Tech are different—like his concerns that conservatives could be targets of censorship and bias. And that could send antitrust efforts in a distinctly new direction on his watch.
Now read the rest of The Algorithm
Deeper Learning
The US Department of Defense is investing in deepfake detection
The Pentagon’s Defense Innovation Unit, a tech accelerator within the military, has awarded its first contract for deepfake detection. Hive AI will receive $2.4 million over two years to help detect AI-generated video, image, and audio content.
Why it matters: As hyperrealistic deepfakes get cheaper and easier to produce, they hurt our ability to tell what’s real. The military’s investment in deepfake detection shows that the problem has national security implications as well. The open question is how accurate these detection tools are, and whether they can keep up with the unrelenting pace at which deepfake generation techniques are improving. Read more from Melissa Heikkilä.
Bits and Bytes
The owner of the LA Times plans to add an AI-powered “bias meter” to its news stories
Patrick Soon-Shiong is building a tool that will allow readers to “press a button and get both sides” of a story. But trying to create an AI model that can somehow provide an objective view of news events is controversial, given that models are biased both by their training data and by fine-tuning methods. (Yahoo)
Google DeepMind’s new AI model is the best yet at weather forecasting
It’s the second AI weather model that Google has launched in just the past few months. But this one’s different: It leaves out traditional physics models and relies on AI methods alone. (MIT Technology Review)
How the Ukraine-Russia war is reshaping the tech sector in Eastern Europe
Startups in Latvia and other nearby countries see the mobilization of Ukraine as a warning and an inspiration. They are now changing consumer products—from scooters to recreational drones—for use on the battlefield. (MIT Technology Review)
How Nvidia’s Jensen Huang is avoiding $8 billion in taxes
Jensen Huang runs Nvidia, the world’s top chipmaker and most valuable company. His wealth has soared during the AI boom, and he has taken advantage of a number of tax dodges “that will enable him to pass on much of his fortune tax free,” according to the New York Times. (The New York Times)
Meta is pursuing nuclear energy for its AI ambitions
Meta wants more of its AI training and development to be powered by nuclear energy, joining the ranks of Amazon and Microsoft. The news comes as many companies in Big Tech struggle to meet their sustainability goals amid the soaring energy demands from AI. (Meta)