Twenty-five years after the September 11 terrorist attacks transformed American national security policy, the surveillance powers introduced during that emergency remain the subject of intense debate. Now artificial intelligence is creating capabilities that lawmakers in 2001 could scarcely have imagined. The question is no longer simply what governments can collect, but what they can discover, predict and decide about their citizens.
INR Report: Based on analysis by Alan Lolaev, visiting researcher at the Laboratory for Political Geography and Contemporary Geopolitics, Higher School of Economics, Moscow, for RT HERE, published 26 October 2026.
The Law That Changed American Surveillance
On 26 October 2001, just 45 days after the September 11 terrorist attacks, US President George W. Bush signed the USA PATRIOT Act into law. Nearly 3,000 people had been killed in attacks that shattered America's sense of security and produced extraordinary political pressure to prevent another catastrophe.
The legislation was introduced during a period of intense national unity, public fear and overwhelming support for stronger counterterrorism measures. President Bush's approval rating reached approximately 86% in late September, while opinion polling indicated that many Americans were prepared to accept restrictions on certain civil liberties in exchange for greater security.
The Patriot Act expanded investigative authorities, facilitated information sharing between intelligence and law enforcement agencies, and broadened access to certain financial, communications and business records. Its supporters argued that existing laws were inadequate for confronting international terrorist networks capable of operating across borders and communications systems.
Critics questioned whether powers introduced under emergency conditions would eventually extend beyond their original purpose. They also warned that surveillance conducted without sufficiently transparent oversight could undermine constitutional protections and expose innocent people to government scrutiny.
Twenty-five years later, the central concern remains unresolved: when governments obtain extraordinary powers during an emergency, what ensures those powers are subsequently limited, reviewed or withdrawn?
From National Emergency to Permanent Security Infrastructure
The Patriot Act did not create American intelligence surveillance. Federal agencies had conducted domestic and foreign intelligence operations for decades, under various statutory and executive authorities.
What changed after September 11 was the political environment in which expanded surveillance became acceptable, alongside the speed with which new legal powers were introduced.
The original legislation permitted broader use of investigative tools and lowered certain barriers to information sharing between agencies. Supporters regarded these changes as essential to identifying terrorist threats before attacks could occur. Civil liberties organisations argued that some provisions were too broad, insufficiently accountable or vulnerable to misuse.
The consequences were not confined to surveillance law. Arab and Muslim Americans, and people perceived to belong to those communities, experienced increased discrimination and hostility in the aftermath of the attacks. Questions emerged about whether security policies were being applied proportionately and whether constitutional protections were being upheld consistently.
The broader institutional lesson is that emergency legislation can outlast the emergency that justified its introduction. Once government agencies acquire new operational capabilities, removing or restricting them may become politically and administratively difficult.
This does not establish that every surveillance measure was unnecessary. It does demonstrate why the continued existence of extraordinary powers requires scrutiny independent of the circumstances in which they were first authorised.
Edward Snowden Exposed the Scale of Surveillance
The debate changed dramatically in 2013 when former National Security Agency contractor Edward Snowden disclosed classified documents concerning American intelligence surveillance programmes.
The disclosures revealed the scale of certain government data-collection activities, including the bulk acquisition of domestic telephone metadata under a legal interpretation associated with Section 215 of the Patriot Act.
Telephone metadata does not ordinarily contain the spoken content of a conversation. However, information about who communicates with whom, when and how frequently can reveal sensitive relationships, patterns of behaviour and associations.
The controversy prompted legal challenges, congressional scrutiny and public debate about whether intelligence agencies had exceeded the privacy expectations of ordinary Americans.
In 2015, Congress enacted the USA FREEDOM Act, which ended the NSA's previous bulk domestic telephone-records collection programme under Section 215 and introduced changes to the legal framework governing access to certain records.
This was a significant restriction on one form of surveillance, but it did not dismantle the wider American intelligence system.
Other legal authorities remained in place, including provisions enacted separately from the Patriot Act.
The Surveillance Powers That Remain
One important distinction in Lolaev's analysis concerns the different legal foundations of American surveillance.
Section 215 of the Patriot Act became associated with the bulk telephone metadata programme revealed by Snowden. Section 702 of the Foreign Intelligence Surveillance Act, enacted in 2008, provides a separate authority for targeting non-US persons reasonably believed to be outside the United States for foreign intelligence purposes.
Under Section 702, communications involving Americans may also be acquired incidentally when they communicate with foreign intelligence targets. The resulting privacy concerns include how collected information is retained, searched and accessed by government agencies.
Executive Order 12333 provides another framework governing substantial elements of American intelligence activity conducted overseas.
These distinctions matter because describing all contemporary American surveillance as a product of the Patriot Act would be legally inaccurate.
The more significant issue is how multiple authorities, technological capabilities and intelligence practices interact to create a surveillance system whose overall reach may be difficult for citizens to understand.
A democratic society may authorise intelligence collection for legitimate national security purposes while still requiring meaningful limits on how information is obtained, retained, analysed and used.
Artificial Intelligence Changes the Equation
The arrival of advanced artificial intelligence introduces a fundamentally different dimension to surveillance.
Traditional intelligence investigations often required analysts to examine records, identify relationships and develop conclusions through substantial human effort. Machine-learning systems can now process enormous quantities of information, identify correlations and generate assessments at speeds that would previously have been impractical.
Communications metadata, financial transactions, location information, online activity and other digital records can potentially be analysed together to reveal patterns that are not apparent when individual datasets are examined separately.
Such capabilities may assist legitimate investigations by identifying previously undetected threats or connections between individuals involved in criminal activity.
However, the same technology creates risks when applied to people who are not suspected of wrongdoing.
Artificial intelligence can identify associations, infer characteristics and assign classifications based on patterns in data. Those outputs may appear authoritative even when the underlying information is incomplete, misleading or biased.
The critical distinction is between identifying a statistical correlation and establishing evidence of unlawful conduct.
An algorithm may identify a person as potentially relevant to an investigation without demonstrating that the person has committed an offence.
If automated assessments influence surveillance decisions, investigations or restrictions on individuals, the consequences of inaccurate conclusions can become substantial.
When Algorithms Become Instruments of State Power
Artificial intelligence also creates accountability problems that traditional surveillance law was not designed to address.
An intelligence agency may collect information under a legally authorised process, but the subsequent analysis of that information can introduce new questions about accuracy, discrimination and procedural fairness.
Some machine-learning systems produce outputs whose reasoning cannot easily be explained, even by the officials operating them. If those outputs influence consequential decisions, affected individuals may struggle to understand why they were identified or how to challenge an error.
Biases present in training data can also be reproduced or amplified. A system trained on historically uneven enforcement or surveillance records may identify particular communities more frequently without establishing that those communities present greater actual risks.
The concern is not that every application of artificial intelligence is inherently unlawful or abusive. It is that increasingly powerful analytical capabilities may develop faster than the legal and institutional safeguards governing their use.
Meaningful oversight therefore requires more than controlling access to raw information. It also requires examination of how information is processed, how algorithmic conclusions are validated and who remains accountable for decisions informed by those conclusions.
The Question of Consent and Democratic Accountability
The original political justification for expanded surveillance after September 11 rested heavily on national security.
Many Americans accepted that extraordinary threats might require additional investigative powers. Yet public support during a national emergency does not necessarily amount to permanent consent for every future use of those powers.
Technological development complicates that question.
A law written to authorise access to particular categories of records may later operate in an environment where those records can be combined with information from commercial databases, digital platforms and automated analytical systems.
The practical consequences of collecting information can therefore change even when the wording of the original law remains substantially unchanged.
This raises a question of democratic accountability: should the expansion of technological capability automatically expand what government agencies can discover about individuals, or should legislatures explicitly reconsider the limits of those powers?
The distinction is fundamental to maintaining effective civilian oversight of intelligence institutions.
Security and Liberty Are Not Mutually Exclusive
Defenders of intelligence surveillance argue that governments have an obligation to identify threats before they result in violence. Terrorism, espionage and organised criminal networks may operate through communications systems that require specialised investigative capabilities.
That argument has genuine force. Governments cannot protect the public effectively if lawful investigative tools are rendered incapable of addressing serious threats.
However, the existence of a legitimate security objective does not eliminate the need to examine whether particular surveillance practices are necessary, proportionate and subject to independent oversight.
Critics argue that broad collection powers, secret legal interpretations and inadequate transparency can undermine public confidence in democratic institutions.
Both concerns require examination through evidence rather than assumptions.
The appropriate questions include whether a surveillance power has demonstrated effectiveness, whether less intrusive alternatives exist, whether independent authorities can review its use and whether individuals have meaningful remedies when mistakes occur.
Artificial intelligence makes these questions more urgent because analytical systems may generate conclusions affecting large numbers of people with relatively little human intervention.
Does This Affect New Zealand?
The American experience has direct relevance to New Zealand because many of the underlying questions also arise within this country's intelligence, policing and digital governance frameworks.
New Zealand's intelligence agencies operate under the Intelligence and Security Act 2017, which establishes statutory functions, authorisation requirements and oversight arrangements. The country also participates in the Five Eyes intelligence partnership alongside the United States, United Kingdom, Australia and Canada.
Participation in international intelligence cooperation raises legitimate questions about how information is shared, what protections apply to New Zealanders and how domestic oversight operates when intelligence is obtained through foreign partners.
The existence of intelligence-sharing arrangements does not itself demonstrate unlawful surveillance. Nevertheless, the international nature of modern intelligence collection makes transparency and accountability particularly important.
New Zealand's Privacy Act 2020 provides protections concerning the collection, use and disclosure of personal information, while other legislation governs intelligence activities and law enforcement access to information.
Artificial intelligence introduces additional challenges because the analytical value of personal data can increase dramatically when separate datasets are combined.
The policy issue is therefore not confined to whether information was collected lawfully. It also concerns what subsequent analysis can reveal, who may access those conclusions and whether individuals retain meaningful protections against inaccurate or unjustified decisions.
For New Zealand, this debate also intersects with digital identity systems, biometric technologies, automated government decision-making and the use of commercially collected information.
These technologies are not necessarily equivalent to intelligence surveillance, but they raise overlapping questions concerning consent, data retention, purpose limitation and independent oversight.
Will this make New Zealand stronger? The relevant test is whether technological development improves public safety and government effectiveness without diminishing the legal protections, privacy rights and democratic accountability that distinguish a free society.
The Real Legacy of the Patriot Act
Twenty-five years after its introduction, the Patriot Act remains an important reference point in the relationship between national security and civil liberties.
Its legacy is not simply that surveillance powers expanded after September 11. It is that an emergency accelerated changes to the legal and institutional environment governing intelligence collection, some of which subsequently required reform.
Snowden's disclosures demonstrated that surveillance programmes could operate on a scale not widely understood by the public. The USA FREEDOM Act demonstrated that legislative restrictions were possible, even after controversial powers had become embedded within national security institutions.
Artificial intelligence now presents a new challenge.
Governments may be able to draw increasingly detailed conclusions about individuals from information that was originally collected for unrelated purposes. The power to analyse information may become as consequential as the power to collect it.
The central question is therefore no longer whether governments should possess any surveillance capability. Modern states already do, and legitimate security functions depend on some of those powers.
The question is who defines their limits, who verifies their effectiveness, who investigates misuse and who is accountable when technology produces harmful or incorrect conclusions.
A quarter-century after the Patriot Act, the balance between security and freedom remains a continuing responsibility of democratic government rather than a decision that can be settled permanently during a national emergency.
Sources
Editorial note: The original RT lead illustration has been retained with its AI-generated image attribution. The report distinguishes established legislation and documented historical developments from analysis concerning the future risks of artificial intelligence in surveillance.