Citizens equipped with powerful artificial intelligence models and automated agents could overwhelm government systems, raising urgent questions about access, security, and public accountability.
The warning centers on a sharp imbalance. AI can help one person create thousands of requests, appeals, complaints, or public comments. Government agencies, however, may still need workers to review each submission.
“Equipped with models and agents, citizens could bring government to a halt.”
The concern is not limited to a single country or agency. Any public body that accepts digital submissions could face pressure from automated activity. That includes tax offices, courts, benefits programs, regulators, and records departments.
Automation Changes the Scale of Participation
Citizens already use formal channels to challenge decisions and demand information. AI agents could make those tools easier to use by drafting documents, tracking deadlines, and filing repeated submissions.
This could benefit people who cannot afford lawyers or specialist advisers. An agent might explain a complex form, prepare an appeal, or identify a missed benefit. Such support could make public services more accessible.
The same tools could also be used to create disruption. A small group might generate large volumes of plausible filings at little cost. Even valid requests could strain an agency if their number rose too quickly.
Several types of government activity may be exposed:
- Public records and information requests
- Permit applications and administrative appeals
- Regulatory comments and formal complaints
- Benefits claims, tax disputes, and court filings
Public Access Versus System Abuse
The central policy challenge is distinguishing lawful participation from coordinated obstruction. Volume alone may not prove abuse. A surge in claims could reflect real public need or a failure within an agency.
Strict limits could protect administrative capacity, but they could also block legitimate users. Identity checks and filing fees may discourage automated attacks while harming low-income residents or people seeking anonymity.
Automated screening presents another risk. If agencies use AI to reject suspected spam, errors could delay benefits, appeals, or legal remedies. People may also struggle to learn why a machine downgraded their submission.
Governments therefore face pressure from both directions. They must prevent systems from being flooded while preserving rights to petition, appeal, and obtain public records.
Agencies May Need New Defenses
A practical response would combine technology, staffing, and clear rules. Agencies could monitor sudden filing spikes, group duplicate material, and prioritize cases involving urgent harm.
They could also require users to confirm key facts rather than banning AI-assisted documents. Rate limits may help in some services, although exceptions would be needed for journalists, researchers, lawyers, and civic groups.
Public agencies may need their own automation to sort submissions and identify repeated arguments. Human review would remain important for decisions affecting rights, money, or legal status.
The larger issue is whether institutions built for human-speed participation can function under machine-speed demand. AI agents may widen access to government, but they may also make disruption cheap and persistent.
Future rules will need to focus on behavior rather than the mere use of AI. The key test will be whether governments can maintain reliable services without weakening lawful public participation.
