The network’s newest member, led by the CEO Sandy Carter, brings message about data ownership to Washington and says that companies who build a “trust architecture” for AI agents will be more profitable for it.
WASHINGTON, D.C. — The Washington AI Network’s newest member arrives with a simple message for the age of AI: As assistants learn more about the people who use them, the most important privacy protection may be the ability to take it back. That is the case Sandy Carter, CEO of EQUS, is bringing to Washington.
“It’s your data. You decide what happens to it,” Carter said. “That’s how you know you really own it, right? When you can delete the data as well.”
Washington AI Network founder Tammy Haddad welcomed EQUS into the network as she spoke with Carter, who has held senior leadership roles at IBM and AWS, for the Washington AI Network Podcast. “The big announcement for me is they’re new members of the Washington AI Network,” Haddad said. Carter said word of the network had reached her through her peers. “All of my other CEO buddies have been telling me about what you guys are doing,” she said. “You’re really having a big impact.”
Carter joined EQUS recently to lead its push to give people control over the data that powers their AI. The company enters the network fresh off the launch of EQUS X, a personal AI built on data the user owns, and with plans to take its case for data ownership to policymakers and the public.
The idea landed with Haddad. “In all the interviews I’ve done and the three years at Washington AI Network, no one said the words ‘delete the data,’” she said.
“Delete, delete, delete”: Taking Your Data Back
Carter traces deletion’s place at the top of EQUS’s priority list to a phone call with a friend diagnosed with cancer who had entered his health information into an AI tool. He soon began seeing ads for counseling and life insurance. “He said, ‘Sandy, all I want to do is delete what I put in. I don’t want it to know that about me,’” she recalled. There was no way to do it.
The stakes rise as people treat AI less like a search engine and more like a confidant. “With most of the AI usage today being AI companions, it’s like your best friend. It knows so much about you,” Carter said. Without a way to go back and erase a passing remark, “you’re going to end up with information about you that gets spread without your approval.”
Deletion may carry a side benefit. When Haddad described an Aspen Ideas Festival session where speakers and audience members alike complained about sycophantic AI, Carter agreed that clearing data from chats and memory would make a significant difference because the system would stop echoing back what it had learned about the user.
Higher stakes than social media
Carter raised her two daughters to be wary of posting photos or anything else that could reveal their friends or whereabouts online. AI, she argues, raises the stakes considerably. “AI is learning the patterns of how you think, of how you negotiate,” she said. “What are the things that make you you? What is your identity? By protecting your data, you don’t release that out into the wild.”
The public, she suggests, wants that protection but doesn’t yet know how to get it. EQUS research found that 79% of people agree they should own and control the information used by personal AI assistants, yet 46.6% have never tried to delete what an AI assistant has saved about them. “There is a desire to own your data,” Carter said. “I think a lot of it is just that people don’t understand how to own their data.”
Who pays when the agent gets it wrong
EQUS describes its roadmap in three words: own, delegate, represent. The second step takes on a question that grows more pressing as AI agents begin spending money and signing agreements for the people who deploy them: who is accountable when they get it wrong?
Carter’s answer is the EQUS Developer Toolkit, which attaches credentials to agents so they act only with the user’s permission and only for a set period. “Just like you have a driver’s license that credentials you, that says, ‘Yes, you’re over 21, you can get in here,’ this credential allows you to buy something on my behalf or your behalf, but it won’t do it unless I give you permission to do it,” she said. “It also has an expiration on it.”
A trust architecture — and why it pays
Carter does not expect any government to slow AI development down. What she does expect is a set of rules for how agents may be used — what she calls a “trust architecture” — spurred by recent high-profile cases of AI agents acting beyond their bounds. “I don’t think it’s slowing down for the sake of slowing down,” she said. “I think it’s putting in a trust architecture so that you’re not concerned or worried.”
Her open question is whether it will take a major catastrophe to get there. Rather than wait, she is pressing companies to be proactive.
She also rejects the idea that guardrails are a drag on the bottom line. Citing The ROI of AI Governance, a paper she co-authored with The Digital Economist, Carter said the evidence points the other way. “We showed through example after example after example that using AI in a responsible way with a trust infrastructure, over the long term, you have higher return on investment than if you didn’t,” she said. “It actually makes you more successful.”
Bragging for women in AI
Carter founded Unstoppable Women of AI in 2022, noting that women hold only about 20% of AI data and research roles. The group trains women online and in person, convenes gatherings at events including CES, Davos and SXSW, and publishes an annual list of the 100 Most Inspirational Women in AI, with the 2026 edition coming soon. “A lot of women don’t brag about themselves,” Carter said, “so I’m like, I can brag for them.”
She is also finishing a new book on data ownership. Her last, AI First, Human Always, shipped with its own AI agent trained on her latest research so readers wouldn’t be left with outdated material — an agent she is now upgrading into a digital twin.
A new voice in the network
For Haddad, EQUS brings a perspective the network’s conversations about AI governance have been missing. “I can’t tell you how happy I am to have you part of the Washington AI Network,” she told Carter. Carter said the company is ready to get to work. “We can’t wait to have a tremendous positive impact,” she said.
The Washington AI Network Podcast episode featuring Sandy Carter is available on YouTube, Apple Podcasts, Spotify and wherever podcasts are found.