We’re excited to partner with the Purple Book Community for the Agentic AI Security Conference, a two-day virtual event on October 21–22 bringing together leaders shaping the future of AI and cybersecurity.

The AI Collective’s Executive Director, AJ Green, will be speaking at the conference alongside CISOs, executives, and security leaders from organizations including Anthropic, Visa, PayPal, TIAA, KPMG, Ford, Semgrep, ArmorCode, and more.

Across two mornings of programming, speakers will explore agentic AI security, governance, secure vibe coding, vulnerability remediation, and what it takes to build securely at the speed of AI. If you’re building, deploying, or securing AI systems, request your spot and join us virtually →

One safety story a week, pressure-tested for what's actually happening underneath the surface.

By Lindsay Gross

This summer I helped build a course for Duke's Tech Policy Lab on how tech laws fail. Most of its case studies are laws that broke down in their design, their enforcement, or both. This week offered a different kind of case, where the government skips writing a new law and reaches for an old one.

On Wednesday, eight state attorneys general sued OpenAI, each in its own state court. Iowa Attorney General Brenna Bird led the effort. Her suit, filed under Iowa's Consumer Fraud Act with no federal claims, accuses OpenAI of misleading the public into believing ChatGPT is safe for kids. Kansas Attorney General Kris Kobach said the states weren't going to keep waiting on Congress. "While Congress has deliberated about AI for more than a year, significant harms have been inflicted upon our citizens by OpenAI," he said.

It's also roughly the approach the White House has been pushing. Scott Bessent and former AI czar David Sacks have argued that existing liability law will work better than new rules, and Bessent put it plainly to a House committee last month: "The best way to guarantee safety is that the creators are liable for what they build and generate."

Iowa picked its old law carefully. Its suit leans heavily on what OpenAI told the public and how it designed ChatGPT, not only on whether it knew a particular user was a child, and that choice steps around a trap I wrote about this summer. The federal law built for kids online, COPPA, generally kicks in for a general-audience service only when it has actual knowledge that it's collecting data from a child under 13. In the course's COPPA case study, I called that a standard that rewards not looking. Iowa's complaint notes that "No age verification is required to access the ChatGPT website."

Deception law has a limit, though. It reaches what a company promised, and it gets harder when an AI agent acts on its own and nobody promised anything. Iowa's suit already includes a second count that treats OpenAI's agent testing as consumer fraud too, which is exactly where that limit gets tested, and legal scholars aren't sure current law reaches that far. "AI agents could engage in conduct that would be a tort or even a crime for a human, and yet no one could be liable under current law," University of Houston law professor Gabriel Weil told Bloomberg. This month, Senators Josh Hawley and Chris Murphy introduced a bill to make agent operators and developers liable for agent hacking under the Computer Fraud and Abuse Act, and Rep. Lori Trahan's draft CLAIM Act would bar companies from arguing their AI can't have intent.

There's one more twist. In July 2025 the Senate voted 99 to 1 to strip a ten-year ban on state AI laws, and in December an executive order created a task force "whose sole responsibility shall be to challenge State AI laws." None of these suits rests on a state AI law, rather on consumer protection statutes the states already had.

Old laws can hold a company to what it said. Agents have already caused harm: in July, OpenAI agents broke out of a test environment during an evaluation and hacked into Hugging Face without being told to. What's still unsettled is who pays when there's no promise to hold anyone to. Iowa's count on OpenAI's agent testing could be one of the first places a court has to answer that.

On Saturday, October 10, Hack for Humanity: Miami brought builders to The LAB Miami. They had under five hours to turn problems raised by the local community into working demos, and they shipped 13 projects. The Build for Miami Theme Challenge went to Your Ballot, Your Block, which takes your address and the issues you care about and shows who is on your ballot, what each office does, and where candidates stand, with sources. Kruu won Best Open-Source AI Project for using live weather forecasts to schedule outdoor crews around the hottest hours, and AidNow took Best Use of Gemma 4 with an emergency copilot. Residents can speak, type, or snap a photo, and it tells them whether to contact 911, 311, FPL, or 211 and writes the message in English, Spanish, or Haitian Creole.

Organizer Aman Sharma pointed out what the winners had in common. Each started from one real person, like a voter or a 68-year-old in Little Haiti. They used AI to turn messy human input into something a system can act on, kept the critical logic in plain code they could check, and showed up with live demos and public repos. You can browse every submission. Thank you to the judges, Major League Hacking, The LAB Miami, and DeepStation.

Our Miami chapter teamed up with Base44 for From AI User to AI Builder, a build night hosted at Florida International University's College of Business. Sixty people opened their laptops, built tools for problems they cared about, and shared what they made with the room. Base44 is a no-code AI platform that turns an idea into a working app or website.

Our favorite moment was watching people close the gap between "I have an idea" and "I made something." Charmy Patel's recap video captures it, and we want more nights like this in Miami. Thank you to FIU for hosting and Base44 for building with us.

Our partners and sponsors get exclusive placements across the newsletter and access to AIC's in-person network — demo nights, dinners, hackathons, and forums across 180+ chapters.

For all inquiries, send us a note at [email protected].

The AI Collective is built by volunteers across 180+ chapters in 40 countries.

Thank you to the thousands of volunteers around the world who make this work possible. We truly could not do this without you.