​A note from Alix | Podcast Highlights | Highlights from the Collective​

The enemy of my enemy is…also my enemy?

Hey everyone,

Anthropic’s PR machine is working overtime to cast the company as AI’s “good guy.”

Just yesterday, we saw Anthropic co-founder Chris Olah cozying up to Pope Leo, nodding along to the Pope’s shallow critiques of AI. The performance of humility stands in sharp contrast to how Anthropic is actually moving through the world.

Last week, we put out an episode with Madeline Batt called “Anthropic is NOT the good guy” on the fault lines between Anthropic and the US Department of War. We also dropped a YouTube short on Anthropic’s compute deal with SpaceX, tied to the Colossus supercomputer cluster in Memphis - a deal that landed just as Tennessee House Republicans moved to strip Black Memphians of democratic representation.

What connects these isn’t timing or coincidence. It’s a pattern where “responsible AI” sits atop very material infrastructure and uneven political geography.

Anthropic has successfully positioned itself as the “good guy” in the public consciousness. In parts of tech policy and philanthropy, that framing has stuck. Omidyar Network and Ford hold stakes, and nonprofit tech conferences often treat it accordingly.

Long term, I think this framing is a mistake.

What Karen Hao calls “clean coal AI” captures the dynamic: this is not a different trajectory, it’s a more palatable story for the same underlying buildout. The result is either quiet co-signing of a system that makes the world worse, or a slower normalization of it under the aesthetics of responsibility - both of which delay any serious reckoning.

It’s enticing to find a “good guy” amid the truly grim landscape of AI companies. But I think we owe it to the people who become collateral damage - the people breathing in Claude-powering noxious fumes, the people losing representation in the face of exploitation - to maintain scrutiny, even when we’re told some actors are better than others. If “clean coal AI” becomes the compromise position, who absorbs the cost?

I’ll be talking more about Anthropic’s bid for moral authority in a livestream next Friday, June 5th, at 12pm ET (register here to join me). If you want to hear directly from people organizing on the ground in Memphis, I sat down with KeShaun Pearson for an interview. Check it out here.

Alix

Podcast Highlights

⏩ Computer Says Kill: A new mini-series exploring the people, politics, and systems that have ushered AI into the business of war

AI is being rapidly integrated into every systematic aspect of our lives under the promise of efficiency and accuracy - and the military is no exception. But that promise is a false one.

The claim that AI can make war cleaner, faster, safer, and more surgical glosses over a harder reality: AI is making war less accurate, less accountable, and more consequential. Integrating AI into warfare doesn’t evade the human costs. It just obscures who is responsible for them.

We’ve seen these promises before - with the bomb, with the drone, and now with AI. But promises are made by people. And so are the decisions that break them. This series traces the people who built the systems, engineered the technology, infiltrated the institutions, compressed the kill chain, and started the wars.

Highlights from the New Protagonist Network

Online harms

• ​GLAAD published their sixth annual Social Media Safety Index, a yearly evaluation of public-facing policies related to LGBTQ safety, privacy, and expression across six major social media platforms. With the exception of TikTok, 2026 platform scores dropped across the board and have hit historic lows.
• Adele Walton spoke to The Guardian, The Telegraph and The Independent about Ofcom’s £950,000 fine against the pro-suicide website that has taken 164 UK lives, what this means for bereaved families and survivors, and what must come next to make sure no one is lost to online suicide harms.

Public sector tech

• ​Priya Goswami’s web-based video game SUPER ASHA, which takes the player into the exhausting daily routine of an Indian health worker, was featured in the Hindustan Times.
• ​Louise Crow’s new piece in Public Technology argues “We need to get better at asking questions about government AI systems.”

Defense and security

• ​FOTI’s new report “Cloud Defense: a European flank” exposes how a majority of EU countries rely on US cloud services for their national defence and security agencies, putting them at risk of a “kill switch.”
• ​Statewatch published the secret agreement negotiated by the European Commission with the Trump administration on the exchange of data for border screening.

Ads & monetisation

• ​What To Fix’s latest investigation shows that Facebook, YouTube, TikTok and X allowed accounts involved in pro-Russia foreign information manipulation to be monetised, despite their owners being under EU sanctions.
• ​Inayat Sabhikhi co-authored “Money from Misery: How Meta Profits from & Exposes Indians to Scams,” a new report for the Forum for Developing Communities, Ekō, and the Bard Human Rights Project, with coverage in The Hindu.

Find out more about the New Protagonist Network and apply to join.

ICYMI: NPN Slack updates & welcoming new members

Welcome to our new members! Karolle Rabarison, Piper Hansen, Anna Zhang, Helena Puig Larrauri. Learn more about them on the #introduce-yourself channel on Slack.

NPN Programming:

• ​June 18: our next Broadcast Media Training, delivered by our partner NEON. Spaces are limited, first-come first-serve.

Some of our conversations on Slack this month:

• ​the cancelation of RightsCon​
• ​this Kashmir Hill piece about teenagers’ use of chatbots​
• the Green Web Foundation’s first "State of the Fossil-Free Internet”
• countering the narrative of inevitability of AI in schools (if you work in education space, reach out to Karolle)