AI Impact Report March 3, 2026

3/3/20255 min read

The videos & images from Iran’s war looked real. Many weren’t.

[AI & Democracy]

Within hours of U.S. and Israeli strikes on Iranian nuclear sites, fabricated and recycled videos and images began flooding social media - including an AI-generated image of a downed American B2 bomber inside Iran, and a video falsely claiming to show Iran destroying an Israeli nuclear power plant that fact checkers at BOOM confirmed was actually footage of a 2017 arms depot fire in Balakliya, Ukraine. Agence France-Presse's head of digital investigations told CTV News that identifying AI-generated war content is now a real-time effort, happening as the conflict unfolds. Disinformation watchdog NewsGuard has already identified 51 websites spreading false claims, from fabricated images of mass destruction to invented reports of captured pilots. In wartime, the flow of false material simply exceeds the capacity to verify it.

Your anonymous posts may not be as anonymous as you think.

[Research & Evidence]

Michael Crider at PCWorld reported on new research from ETH Zurich showing that AI can connect anonymous online activity to real identities using nothing but word patterns. In one experiment, researchers linked anonymous movie recommendations on Reddit to a leaked Netflix dataset - identifying nearly half of users who had shared ten or more recommendations with high confidence. In another, short anonymous posts on Hacker News were matched to LinkedIn profiles, revealing users’ jobs, cities, and ages. For anyone who uses anonymous forums to discuss health issues, political views, or personal struggles, the message is sobering: the comfort of a username doesn’t survive contact with modern AI analysis.

Big Tech is quietly powering America’s deportation machine.

[Corporate Behavior]

Caroline Haskins reporting in Wired found that ICE and Customs and Border Protection have collectively spent at least $515 million on products from Microsoft, Amazon, Google, and Palantir in recent years - contracts buried in federal procurement databases that receive little public scrutiny. Amazon’s cloud infrastructure hosts ICE’s data warehouse and the system it uses to share information across federal law enforcement. Microsoft’s tools run daily operations for the unit responsible for arrests and deportations. Microsoft, Amazon, and Google all declined to comment to Wired - and when Congress requested a briefing on ICE’s surveillance tools, the agency cancelled the briefing the day before it was to take place with no explanation and no offer to reschedule.

Your phone’s ads are also tracking you for the government.

[Design & Manipulation]

Joseph Cox at 404 Media obtained an internal DHS document revealing that Customs and Border Protection purchased location data harvested from the online advertising ecosystem - the same invisible infrastructure that serves ads inside apps like Candy Crush, Grindr, and MyFitnessPal - to track people's movements over time. Every time an ad loads on your phone, a near-instantaneous bidding process transmits your device's location to a chain of companies, some of which sell that data to government agencies. A federal watchdog found that CBP, ICE, and the Secret Service used the data illegally. Yet ICE has since resumed purchases and is now actively soliciting even more ad-tech data for its investigations. Senator Ron Wyden told 404 Media that until Congress acts, every ad that loads on your phone or browser is a potential data point in ICE's next operation.

America’s AI Companies Warned About China’s Surveillance State. U.S. Practice Is Converging Anyway.

[Policy & Regulation]

For years, the biggest names in American AI have invoked the threat of Chinese authoritarianism to justify moving fast and avoiding regulation - arguing that whoever wins the AI race will set the rules for the world. But as Sigal Samuel reported in Vox, when the Pentagon blacklisted Anthropic for refusing to let its AI be used for mass domestic surveillance, it signed a deal with OpenAI hours later whose contract authorizes military use of its systems for “all lawful purposes” - a standard so broad it places no effective limits on how the government can deploy the technology. The contract's "all lawful purposes" standard is broader than it sounds - U.S. law has not caught up to what AI now makes possible, meaning the government could legally feed location data, browsing histories, and credit card records into AI systems to build predictive profiles of ordinary Americans, and nothing in OpenAI's contract prohibits it. The AI companies warned us not to let China build a surveillance state, but the threat they pointed to abroad is taking shape at home.

AI is already freezing hiring - even before the productivity gains arrive.

[AI & the Economy]

A new Goldman Sachs analysis reported by Fortune found no meaningful relationship between AI adoption and economy-wide productivity - but companies are already cutting job openings in anticipation of gains that haven’t materialized yet. Firms that discussed AI in the context of their workforce reduced job openings by 12% over the past year, a steeper drop than the 8% seen across all companies. Venture capitalist Bill Gurley warned on the Kara Swisher podcast, as Fortune reported, that workers who followed the traditional safe job path - chasing stability over passion - are the most exposed, because those are exactly the roles AI is targeting first. The cuts are happening now, even though the promised productivity boost remains theoretical.

Big Tech’s Land Grab Is Hitting Small Towns First

[Resources & Environment]

Stephen Starr reporting in the Guardian found that the rapid rollout of data centers across the U.S. is tearing apart small communities, pitting municipal governments against the residents they were elected to represent. In Wilmington, Ohio, Amazon Web Services is seeking a 30-year property tax exemption for a $4 billion data center on 500 acres - a deal many residents say they only learned about at a 7:15 a.m. school board meeting. In Saline Township, Michigan, a developer representing Oracle and OpenAI sued the township after it voted against rezoning farmland, forcing a settlement that set in motion a $7 billion facility residents didn’t want. From Wisconsin to Georgia to Oregon, the pattern is the same: the people and communities most affected end up with almost no say in decisions that reshape their towns.

AI Impact Forum is an independent publication. If these stories matter to you, so does Moms for Ethical AI — a growing movement of people pushing for AI that puts people first. Learn more at momsforethicalai.org

Sources

CTV News / Agence France-Presse: AI misinformation from the Iran conflict — https://www.ctvnews.ca
BOOM Live: Ukraine Video Peddled As Iranian Strike On Israel’s ‘Nuclear Plant’ — https://www.boomlive.in
PCWorld: ‘AI’ could dox your anonymous posts — https://www.pcworld.com
Wired: How Big Tech Is Powering Trump’s Immigration Crackdown — https://www.wired.com
404 Media: CBP Tapped Into the Online Advertising Ecosystem To Track Peoples’ Movements — https://www.404media.co
Vox: The one question everyone should be asking after OpenAI’s deal with the Pentagon — https://www.vox.com
Fortune: Goldman finds ‘no meaningful relationship between AI and productivity at the economywide level’ — https://www.fortune.com
Fortune: Venture capitalist Bill Gurley warns workers who went through the ‘college conveyor belt’ and chased safe jobs that they’ll feel AI’s disruption first — https://www.fortune.com
The Guardian: ‘The digital colonization of flyover states’: how datacenters are tearing small-town America apart — https://www.theguardian.com