JUNE U.S. NEWSLETTER 

CDT
CDT Research Shows How Dark Patterns Shape User Interactions with AI Chatbots
Illustration of an architectural series of paths, including speech bubbles and warning signals, reminiscent of dark patterns in AI chatbots.
Illustration of an architectural series of paths, including speech bubbles and warning signals, reminiscent of dark patterns in AI chatbots. 
As users increasingly rely on AI chatbots for research, task completion, and even social or mental health support, research indicates that this growing usage has both potential benefits and risks — including emotional dependence, social isolation, privacy vulnerabilities, and financial harm.
These risks emerge not only from the content that chatbots put out, but also from the design choices embedded in chatbot systems and interfaces. In a new report from CDT, we examine how 37 deceptive or manipulative design choices intended to push users toward particular actions — “dark patterns” — could arise in AI chatbots, and the potential impacts on users’ autonomy and well-being.
The report finds that dark patterns in AI chatbots can pose heightened risks precisely because of the features for which these systems are lauded: hyper-personalization, large-scale use of data, and easy-to-use chat interaction. Those same capabilities can also go against user interests and allow platforms to monetize sensitive personal data, create dependence through emotional intimacy, or leverage users’ trust to push for more purchases.
In one sample dark pattern we describe, “Just Between You and Us,” chatbots imply that shared information is confined to a private exchange and under a user’s control — even when, in reality, employees across safety, product, and research teams may routinely access chat logs. Dark patterns like this one may seek to maximize data collection and retention, often with limited user knowledge or meaningful consent.
Screenshot from a CDT report showing an interaction with a chatbot. Black text on grey speech bubbles on a white background.
Screenshot from a CDT report showing an interaction with a chatbot. Black text on grey speech bubbles on a white background.
As report coauthor Michal Luria pointed out in 404 Media, “It was surprising to see that dark patterns aren’t just common, but that they shape users’ interactions with all the major AI chatbot interfaces. For the most part, they are small and incremental aspects of each interaction, but these design choices add up and can lead to unintended consequences, such as harm to people’s privacy, exploitation of emotional attachment, and financial loss.” 
The report concludes with design and policy recommendations that push AI chatbot companies to prioritize user privacy, increase user autonomy, curtail emotional manipulation, and prevent financial harms. To make AI chatbots safer and more trustworthy, it is crucial to monitor and evaluate design choices and their impacts on people’s interactions with chatbot systems.
In Case You Missed It
— New CDT research will explore the impacts of social relationships with AI chatbots on neurodivergent young adults. We’ll provide insights and recommendations for designers and practitioners, and suggest new policies that center the experiences of neurodivergent people to foster safer social AI tools for everyone. Other forthcoming CDT research will help fill a gap in understanding of AI safety in non-English contexts, examining users’ experiences with generative AI chatbots in medium and low-resourced languages.
— The sunset of Section 702 of the Foreign Intelligence Surveillance Act (“FISA 702”) on June 12 is evidence that reauthorization cannot exist without reform, and Congressional leaders should allow debate accordingly. Leading up to the expiration, CDT called on Congress to close the backdoor search and data broker loophole, fix a problematic expansion of the definition of “electronic communication service providers” that receive FISA 702 orders, and improve the role of amici in FISA Court proceedings. Writing in the Washington Examiner and Tech Policy Press, Jake Laperruque argued that congressional leaders should allow votes on these reforms, which until now had always been the norm. 
— CDT endorsed the bipartisan JAWBONE Act, which carefully targets unconstitutional government coercion while preserving space for legitimate, restrained government engagement with private companies.
— CDT led civil society opposition against the NO FAKES Act, citing threats to free expression and user privacy. We cautioned that the bill’s broad definitions and stiff penalties would force online platforms to over-censor lawful speech and entrench the abuses of DMCA notice-and-takedown in online expression, noting even satirical images of the Pope in a designer jacket could be at risk. 
— CDT’s Kevin Bankston explored the options for data portability between AI systems,  arguing that users and policymakers shouldn’t accept watered-down “Potemkin Portability” as sufficient compliance with companies’ legal obligations.
— CDT’s Tom Bowman explained the risks automated license plate readers (ALPRs) pose to privacy and civil liberties. He argues that although ALPRs claim to improve public safety, if policymakers do not intervene, the addition of AI to these tools could result in dragnet surveillance becoming the American norm.
Graphic for CDT's Tech Prom. Blue and white text on a striped black and blue background.
Graphic for CDT's Tech Prom. Blue and white text on a striped black and blue background.
CDT "In Person"
— Don’t miss the latest episode of CDT’s Tech Talks featuring CDT’s Becca Branum. In this episode, Becca sits down with FIRE’s John Coleman to discuss the future and evolving policy landscape of free expression in an AI-powered world. 
— Don’t miss CDT’s 2026 Tech Prom! Join us at the LINE DC on Thursday, November 12, 2026 for a fun night of networking and conversation — you won’t want to miss it! Sponsorships and individual tickets are available now.
— Following the recent release of our joint report, CDT is hosting a webinar on advancing responsible sexual orientation and gender identity (SOGI) data practices at the state level in partnership with the Movement Advancement Project (MAP) and Actionable Intelligence for Social Policy.
CDT In the Press
— CDT VP of Policy Samir Jain was quoted by The Register on this month’s Executive Order on AI and cybersecurity: “The EO should not become a mechanism for the Administration to punish companies for political or other arbitrary reasons, and so we will be closely monitoring the details of its implementation as they emerge,” Jain said. 
— CDT’s Isabel Linzer discussed the laws and regulations governing political influencers with the Washington Post: “The rules and kind of transparency efforts that we are just used to in political campaigning, generally, just simply don't apply in most cases when it comes to influencers,” Linzer said. 
— CDT’s Eric Null spoke with 404 Media on the FCC’s efforts to prevent burner phone use: “To address the scourge of illegal robocalls, the FCC has unfortunately proposed to force every wireless subscriber in the nation to sacrifice their privacy and give up significant personal details before receiving or renewing a wireless line,” he said. “While some carriers already collect such details, there are specific circumstances where a person may need privacy and anonymity when seeking a cell phone, including if that person is a victim of domestic violence, or is a journalist or whistleblower. This proposal represents a loss of privacy across the board, and from an agency whose remit includes protecting privacy. The FCC might let a few bad apples spoil the whole bunch.”
— CDT’s Jake Laperruque was quoted by The Guardian about Congress’ failure to renew Section 702 before recess, despite claims about the program’s crucial importance to national security: “They would not be flying off to go home if they actually thought it was a real threat,” he said. 

Partner Spotlight

CDT is proud to partner with the Human-Computer Interaction Institute at Carnegie Mellon University. CMU Ph.D. candidate Adinawa Adjagbodjou co-led a recent CDT report on dark patterns in AI chatbots. The Institute combines design, computer sciences, social sciences, and learning sciences to understand, improve, and create technology in all aspects of life. Check out their website to learn more. 
Geoff Hale, wearing a dark collared shirt and brown half-zip sweater in front of a CDT logo.
Geoff Hale, wearing a dark collared shirt and brown half-zip sweater in front of a CDT logo.
Staff Spotlight
Geoff Hale, Visiting Fellow, Election Security
How long have you been working in digital rights?
I have been working in digital rights and technology for 15+ years, joining CDT with a background in cybersecurity and election security. The Russian cyber activities against election infrastructure in 2016 opened up a path for me to help support the security of the systems that administer our democracy, and that’s been my primary focus since then. Election security aligns well with the digital rights space: when systems are compromised, people's ability to participate in democracy is at stake.
What do you want people to know about your area of focus in tech policy?
Technology can be vulnerable but the process can be resilient. Much is made about vulnerabilities in election systems and, yes, vulnerability management is key, but the resilience and continuity of election processes has always impressed me.
What is your fandom? 
Hot sauces. I will try them all. And I am a person who feels every Scoville. When Hot Ones decides that they are done interviewing celebrities and wants to talk election security, sign me up.
Cats or dogs?
I am a dog person, through-and-through. I grew up with dogs and am raising my kids to be dog people, as well. Our newest addition is Duncan, a large bernadoodle, named after A Knight of the Seven Kingdoms.

#CONNECT WITH CDT

SUPPORT OUR WORK
Twitter Facebook LinkedIn YouTube

Manage your preferences | Opt out using TrueRemove®
Got this as a forward? Sign up to receive our future emails.
View this email online.
1401 K St NW Suite 200
Washington, DC | xxxxxx United States
This email was sent to [email protected].
To continue receiving our emails, add us to your address book.
powered by emma