Research program

Investigating whether AI can strengthen human agency online.

Six workstreams spanning behavioral science, human–computer interaction, privacy-preserving machine learning, ethics, and mobile operating-system design.

Main research question

How can AI-enabled mobile devices help people exercise meaningful control over their own connectivity during periods when their immediate impulses may conflict with their previously stated intentions?

Workstreams

Six streams of inquiry

01 · Digital Vulnerability

Understand when and why individuals experience difficulty maintaining intended digital boundaries.

02 · Existing Digital Controls

Study blockers, filters, screen-time systems, parental controls, and other interventions.

03 · Behavioral Pre-Commitment

Study whether decisions made before vulnerability are more durable than decisions made during it.

04 · AI Personalization

Explore how privacy-preserving AI can recommend useful boundaries without invasive data collection.

05 · Ethics & Human Agency

Study consent, autonomy, coercion, privacy, accessibility, bias, and potential misuse.

06 · Responsible OS Integration

Study technical requirements for implementing voluntary connectivity controls at the mobile OS level.

Open questions

Research Questions

These are the questions the program exists to answer. None of them is settled.

  1. 01When do people experience difficulty maintaining self-selected digital boundaries?
  2. 02Why do users disable or circumvent existing blockers?
  3. 03How effective is voluntary digital pre-commitment?
  4. 04What level of restriction do users consider acceptable?
  5. 05Can AI identify recurring risk windows without invasive surveillance?
  6. 06Which behavioral signals are appropriate to use?
  7. 07Which signals should never be collected?
  8. 08Can AI recommendations improve schedule selection?
  9. 09How should recommendations be explained?
  10. 10What happens when AI makes an incorrect prediction?
  11. 11Should users always be able to override SafeTime?
  12. 12Should overrides have cooldown periods?
  13. 13What emergency exceptions are necessary?
  14. 14What should remain available during full SafeTime?
  15. 15How should SafeTime work differently for adults and adolescents?
  16. 16What risks arise from OS-level restrictions?
  17. 17Could this technology be misused?
  18. 18What safeguards would prevent coercive use?
  19. 19How should SafeTime measure effectiveness?
  20. 20Can SafeTime work across different cultures?
  21. 21Could pre-commitment improve users' perceived digital agency?
Flagship study

SafeTime Digital Agency Study 2026

Goal: understand how individuals currently manage unwanted online behaviors and whether voluntary pre-commitment could improve digital agency.

Topics covered

  • Internet habits
  • Unwanted digital behavior
  • Times of vulnerability
  • Existing blocker use
  • Why blockers are disabled
  • User attitudes toward pre-commitment
  • Preferred restriction levels
  • Cooldown preferences
  • Trust in AI
  • Privacy concerns
  • Emergency exceptions
  • Attitudes toward OS integration
Planned target

100–300

Initial exploratory sample

Planned target

500–1,000+

Future expanded study

These are planned targets, not achieved recruitment figures.

Community co-design

Nothing About Users Without Users.

People affected by unwanted digital behavior should participate in designing the tools intended to help them.

Interviews

One-to-one conversations about real digital boundaries.

Surveys

Structured, anonymous, and privacy-respecting.

Focus groups

Group discussion of trade-offs and acceptability.

Prototype testing

Hands-on evaluation of scheduling and cool-off design.

Participatory design workshops

Co-designing features with the people who would use them.

Community advisory sessions

Ongoing guidance from affected communities.

Impact framework

What Would Success Look Like?

Proposed evaluation indicators for future studies.

Increased adherence to self-selected SafeTime sessions
Reduced voluntary override attempts
Increased user-reported sense of control
Reduced unwanted late-night usage
Improved sleep-related digital boundaries
Higher alignment between user intentions and behavior
Trust in AI recommendations
User satisfaction
Privacy confidence
Successful emergency access
Continued voluntary use

These are proposed evaluation indicators and should not be interpreted as demonstrated SafeTime outcomes.

Publications

Research outputs

Working Papers
Research Briefs
Survey Reports
Technical Papers
Policy Briefs
Conference Presentations
Peer-Reviewed Publications

Research publications forthcoming.

  • We publish findings that do not support our hypothesis as readily as those that do.
  • Study protocols will be shared before results where feasible.