But that's not a reason to look away from how they are being used now. The key is to address how AI tools compound harm and ensure they can also provide real support while we fight for broader change. We can, and must, do both.

I got to reunite with @saltzshaker.bsky.social to develop this work, and as we write in this piece in @techpolicypress.bsky.social AI chatbots are NOT the solution to systemic crises rooted in inequality and inaccessibility.

Questions on the table = urgent + unresolved:
→Should chatbots proactively continue conversations with users in crisis?
→ Should they escalate beyond providing a hotline link?
→And more philosophically, what is the role of a technology company in the emotional and informational life of its users?

We were coming together to develop initial, shared guidance for how AI chatbots should handle high-stakes mental health interactions.

This rounds out the *19*-case collection developed over the past two years: a massive effort to better understand opportunities + mitigate the risks of synthetic media.

🔗 Read the full collection here!

syntheticmedia.partnershiponai.org#case_studies

3. How transparency signals can empower users to make informed decisions, from #Google

🔗 Read Here:
partnershiponai.org/google-frame...

2. How disclosure can limit misleading and gendered content, from #Meedan

🔗 Read Here:
partnershiponai.org/meedan-frame...

1. How synthetic media impacts elections and political content, from @code4africa.bsky.social

🔗 Read Here:
partnershiponai.org/codeforafric...

Plus, networking breaks & meet-ups! Follow #ISOJ2025 for more updates! 📲🔥
https://bit.ly/3C7f5Mh

Read the full paper here: www.arxiv.org/abs/2502.04526

And check out the important implications for how we approach AI governance broadly!

#AIPolicy #AIGovernance #deepfakes #syntheticmedia #contentauthenticity

Cross-sector collaboration works when it:

-Complements (not replaces) government regulation
-Has adequate lead time
-Builds on trusted relationships
-Combines social & technical expertise

(One day I'll write a whole playbook on this!)

On technical solutions: AI labels & transparency measures? Important but not sufficient. They need to go beyond simple "AI or not" binaries to be truly useful.

Trust emerged as the foundation.

Not just among policymakers, companies & civil society - but crucially, between these groups & the public they serve. Without trust, even good policies struggle.

Time shapes everything.

Stakeholders draw on past experiences, weigh present contexts, & consider future implications when crafting AI policies. History & foresight matter as much as current tech.

First key insight: Synthetic media isn't just a tech problem.

It's fundamentally about misrepresentation & impersonation. This reframing changes how we approach solutions.

And as the AI agents hype shows no signs of abating, let's not forget the importance of avoiding centralization of power in "personhood" solutions/supporting user choice.

Check out the other incredible papers selected here: fpf.org/press-releas...

- honored to be in such thoughtful company, amongst my (many!) co-authors, and other award winners!